20210518.md

ArXiv cs.CV --Tue, 18 May 2021

1.Divide and Contrast: Self-supervised Learning from Uncurated Data ⬇️

Self-supervised learning holds promise in leveraging large amounts of unlabeled data, however much of its progress has thus far been limited to highly curated pre-training data such as ImageNet. We explore the effects of contrastive learning from larger, less-curated image datasets such as YFCC, and find there is indeed a large difference in the resulting representation quality. We hypothesize that this curation gap is due to a shift in the distribution of image classes -- which is more diverse and heavy-tailed -- resulting in less relevant negative samples to learn from. We test this hypothesis with a new approach, Divide and Contrast (DnC), which alternates between contrastive learning and clustering-based hard negative mining. When pretrained on less curated datasets, DnC greatly improves the performance of self-supervised learning on downstream tasks, while remaining competitive with the current state-of-the-art on curated datasets.

2.The Boombox: Visual Reconstruction from Acoustic Vibrations ⬇️

We introduce The Boombox, a container that uses acoustic vibrations to reconstruct an image of its inside contents. When an object interacts with the container, they produce small acoustic vibrations. The exact vibration characteristics depend on the physical properties of the box and the object. We demonstrate how to use this incidental signal in order to predict visual structure. After learning, our approach remains effective even when a camera cannot view inside the box. Although we use low-cost and low-power contact microphones to detect the vibrations, our results show that learning from multi-modal data enables us to transform cheap acoustic sensors into rich visual sensors. Due to the ubiquity of containers, we believe integrating perception capabilities into them will enable new applications in human-computer interaction and robotics. Our project website is at: this http URL

3.A Light Stage on Every Desk ⬇️

Every time you sit in front of a TV or monitor, your face is actively illuminated by time-varying patterns of light. This paper proposes to use this time-varying illumination for synthetic relighting of your face with any new illumination condition. In doing so, we take inspiration from the light stage work of Debevec et al., who first demonstrated the ability to relight people captured in a controlled lighting environment. Whereas existing light stages require expensive, room-scale spherical capture gantries and exist in only a few labs in the world, we demonstrate how to acquire useful data from a normal TV or desktop monitor. Instead of subjecting the user to uncomfortable rapidly flashing light patterns, we operate on images of the user watching a YouTube video or other standard content. We train a deep network on images plus monitor patterns of a given user and learn to predict images of that user under any target illumination (monitor pattern). Experimental evaluation shows that our method produces realistic relighting results. Video results are available at this http URL.

4.StrobeNet: Category-Level Multiview Reconstruction of Articulated Objects ⬇️

We present StrobeNet, a method for category-level 3D reconstruction of articulating objects from one or more unposed RGB images. Reconstructing general articulating object categories % has important applications, but is challenging since objects can have wide variation in shape, articulation, appearance and topology. We address this by building on the idea of category-level articulation canonicalization -- mapping observations to a canonical articulation which enables correspondence-free multiview aggregation. Our end-to-end trainable neural network estimates feature-enriched canonical 3D point clouds, articulation joints, and part segmentation from one or more unposed images of an object. These intermediate estimates are used to generate a final implicit 3D reconstruction.Our approach reconstructs objects even when they are observed in different articulations in images with large baselines, and animation of reconstructed shapes. Quantitative and qualitative evaluations on different object categories show that our method is able to achieve high reconstruction accuracy, especially as more views are added.

5.Learning to Automatically Catch Potholes in Worldwide Road Scene Images ⬇️

Among several road hazards that are present in any paved way in the world, potholes are one of the most annoying and also involving higher maintenance costs. There exists an increasing interest on the automated detection of these hazards enabled by technological and research progress. Our research work tackled the challenge of pothole detection from images of real world road scenes. The main novelty resides on the application of the latest progress in AI to learn the visual appearance of potholes. We built a large dataset of images with pothole annotations. They contained road scenes from different cities in the world, taken with different cameras, vehicles and viewpoints under varied environmental conditions. Then, we fine-tuned four different object detection models based on Faster R-CNN and SSD deep neural networks. We achieved high average precision and the pothole detector was tested on the Nvidia DrivePX2 platform with GPGPU capability, which can be embedded on vehicles. Moreover, it was deployed on a real vehicle to notify the detected potholes to a given IoT platform as part of AUTOPILOT H2020 project.

6.Unknown-box Approximation to Improve Optical Character Recognition Performance ⬇️

Optical character recognition (OCR) is a widely used pattern recognition application in numerous domains. There are several feature-rich, general-purpose OCR solutions available for consumers, which can provide moderate to excellent accuracy levels. However, accuracy can diminish with difficult and uncommon document domains. Preprocessing of document images can be used to minimize the effect of domain shift. In this paper, a novel approach is presented for creating a customized preprocessor for a given OCR engine. Unlike the previous OCR agnostic preprocessing techniques, the proposed approach approximates the gradient of a particular OCR engine to train a preprocessor module. Experiments with two datasets and two OCR engines show that the presented preprocessor is able to improve the accuracy of the OCR up to 46% from the baseline by applying pixel-level manipulations to the document image. The implementation of the proposed method and the enhanced public datasets are available for download.

7.Pseudo-Label Ensemble-based Semi-supervised Learning for Handling Noisy Soiling Segmentation Annotations ⬇️

Manual annotation of soiling on surround view cameras is a very challenging and expensive task. The unclear boundary for various soiling categories like water drops or mud particles usually results in a large variance in the annotation quality. As a result, the models trained on such poorly annotated data are far from being optimal. In this paper, we focus on handling such noisy annotations via pseudo-label driven ensemble model which allow us to quickly spot problematic annotations and in most cases also sufficiently fixing them. We train a soiling segmentation model on both noisy and refined labels and demonstrate significant improvements using the refined annotations. It also illustrates that it is possible to effectively refine lower cost coarse annotations.

8.Rethinking the Design Principles of Robust Vision Transformer ⬇️

Recent advances on Vision Transformers (ViT) have shown that self-attention-based networks, which take advantage of long-range dependencies modeling ability, surpassed traditional convolution neural networks (CNNs) in most vision tasks. To further expand the applicability for computer vision, many improved variants are proposed to re-design the Transformer architecture by considering the superiority of CNNs, i.e., locality, translation invariance, for better performance. However, these methods only consider the standard accuracy or computation cost of the model. In this paper, we rethink the design principles of ViTs based on the robustness. We found some design components greatly harm the robustness and generalization ability of ViTs while some others are beneficial. By combining the robust design components, we propose Robust Vision Transformer (RVT). RVT is a new vision transformer, which has superior performance and strong robustness. We further propose two new plug-and-play techniques called position-aware attention rescaling and patch-wise augmentation to train our RVT. The experimental results on ImageNet and six robustness benchmarks show the advanced robustness and generalization ability of RVT compared with previous Transformers and state-of-the-art CNNs. Our RVT-S* also achieves Top-1 rank on multiple robustness leaderboards including ImageNet-C and ImageNet-Sketch. The code will be available at this https URL.

9.BigEarthNet-MM: A Large Scale Multi-Modal Multi-Label Benchmark Archive for Remote Sensing Image Classification and Retrieval ⬇️

This paper presents the multi-modal BigEarthNet (BigEarthNet-MM) benchmark archive made up of 590,326 pairs of Sentinel-1 and Sentinel-2 image patches to support the deep learning (DL) studies in multi-modal multi-label remote sensing (RS) image retrieval and classification. Each pair of patches in BigEarthNet-MM is annotated with multi-labels provided by the CORINE Land Cover (CLC) map of 2018 based on its thematically most detailed Level-3 class nomenclature. Our initial research demonstrates that some CLC classes are challenging to be accurately described by only considering (single-date) BigEarthNet-MM images. In this paper, we also introduce an alternative class-nomenclature as an evolution of the original CLC labels to address this problem. This is achieved by interpreting and arranging the CLC Level-3 nomenclature based on the properties of BigEarthNet-MM images in a new nomenclature of 19 classes. In our experiments, we show the potential of BigEarthNet-MM for multi-modal multi-label image retrieval and classification problems by considering several state-of-the-art DL models. We also demonstrate that the DL models trained from scratch on BigEarthNet-MM outperform those pre-trained on ImageNet, especially in relation to some complex classes, including agriculture and other vegetated and natural environments. We make all the data and the DL models publicly available at this https URL, offering an important resource to support studies on multi-modal image scene classification and retrieval problems in RS.

10.Large-Scale Unsupervised Person Re-Identification with Contrastive Learning ⬇️

Existing public person Re-Identification~(ReID) datasets are small in modern terms because of labeling difficulty. Although unlabeled surveillance video is abundant and relatively easy to obtain, it is unclear how to leverage these footage to learn meaningful ReID representations. In particular, most existing unsupervised and domain adaptation ReID methods utilize only the public datasets in their experiments, with labels removed. In addition, due to small data sizes, these methods usually rely on fine tuning by the unlabeled training data in the testing domain to achieve good performance. Inspired by the recent progress of large-scale self-supervised image classification using contrastive learning, we propose to learn ReID representation from large-scale unlabeled surveillance video alone. Assisted by off-the-shelf pedestrian detection tools, we apply the contrastive loss at both the image and the tracklet levels. Together with a principal component analysis step using camera labels freely available, our evaluation using a large-scale unlabeled dataset shows far superior performance among unsupervised methods that do not use any training data in the testing domain. Furthermore, the accuracy improves with the data size and therefore our method has great potential with even larger and more diversified datasets.

11.Multi-object Tracking with Tracked Object Bounding Box Association ⬇️

The CenterTrack tracking algorithm achieves state-of-the-art tracking performance using a simple detection model and single-frame spatial offsets to localize objects and predict their associations in a single network. However, this joint detection and tracking method still suffers from high identity switches due to the inferior association method. To reduce the high number of identity switches and improve the tracking accuracy, in this paper, we propose to incorporate a simple tracked object bounding box and overlapping prediction based on the current frame onto the CenterTrack algorithm. Specifically, we propose an Intersection over Union (IOU) distance cost matrix in the association step instead of simple point displacement distance. We evaluate our proposed tracker on the MOT17 test dataset, showing that our proposed method can reduce identity switches significantly by 22.6% and obtain a notable improvement of 1.5% in IDF1 compared to the original CenterTrack's under the same tracklet lifetime. The source code is released at this https URL.

12.Learning to Relate Depth and Semantics for Unsupervised Domain Adaptation ⬇️

We present an approach for encoding visual task relationships to improve model performance in an Unsupervised Domain Adaptation (UDA) setting. Semantic segmentation and monocular depth estimation are shown to be complementary tasks; in a multi-task learning setting, a proper encoding of their relationships can further improve performance on both tasks. Motivated by this observation, we propose a novel Cross-Task Relation Layer (CTRL), which encodes task dependencies between the semantic and depth predictions. To capture the cross-task relationships, we propose a neural network architecture that contains task-specific and cross-task refinement heads. Furthermore, we propose an Iterative Self-Learning (ISL) training scheme, which exploits semantic pseudo-labels to provide extra supervision on the target domain. We experimentally observe improvements in both tasks' performance because the complementary information present in these tasks is better captured. Specifically, we show that: (1) our approach improves performance on all tasks when they are complementary and mutually dependent; (2) the CTRL helps to improve both semantic segmentation and depth estimation tasks performance in the challenging UDA setting; (3) the proposed ISL training scheme further improves the semantic segmentation performance. The implementation is available at this https URL.

13.Multi-modal Visual Place Recognition in Dynamics-Invariant Perception Space ⬇️

Visual place recognition is one of the essential and challenging problems in the fields of robotics. In this letter, we for the first time explore the use of multi-modal fusion of semantic and visual modalities in dynamics-invariant space to improve place recognition in dynamic environments. We achieve this by first designing a novel deep learning architecture to generate the static semantic segmentation and recover the static image directly from the corresponding dynamic image. We then innovatively leverage the spatial-pyramid-matching model to encode the static semantic segmentation into feature vectors. In parallel, the static image is encoded using the popular Bag-of-words model. On the basis of the above multi-modal features, we finally measure the similarity between the query image and target landmark by the joint similarity of their semantic and visual codes. Extensive experiments demonstrate the effectiveness and robustness of the proposed approach for place recognition in dynamic environments.

14.STRIDE : Scene Text Recognition In-Device ⬇️

Optical Character Recognition (OCR) systems have been widely used in various applications for extracting semantic information from images. To give the user more control over their privacy, an on-device solution is needed. The current state-of-the-art models are too heavy and complex to be deployed on-device. We develop an efficient lightweight scene text recognition (STR) system, which has only 0.88M parameters and performs real-time text recognition. Attention modules tend to boost the accuracy of STR networks but are generally slow and not optimized for device inference. So, we propose the use of convolution attention modules to the text recognition networks, which aims to provide channel and spatial attention information to the LSTM module by adding very minimal computational cost. It boosts our word accuracy on ICDAR 13 dataset by almost 2%. We also introduce a novel orientation classifier module, to support the simultaneous recognition of both horizontal and vertical text. The proposed model surpasses on-device metrics of inference time and memory footprint and achieves comparable accuracy when compared to the leading commercial and other open-source OCR engines. We deploy the system on-device with an inference speed of 2.44 ms per word on the Exynos 990 chipset device and achieve an accuracy of 88.4% on ICDAR-13 dataset.

15.Temporal Prediction and Evaluation of Brassica Growth in the Field using Conditional Generative Adversarial Networks ⬇️

Farmers frequently assess plant growth and performance as basis for making decisions when to take action in the field, such as fertilization, weed control, or harvesting. The prediction of plant growth is a major challenge, as it is affected by numerous and highly variable environmental factors. This paper proposes a novel monitoring approach that comprises high-throughput imaging sensor measurements and their automatic analysis to predict future plant growth. Our approach's core is a novel machine learning-based growth model based on conditional generative adversarial networks, which is able to predict the future appearance of individual plants. In experiments with RGB time-series images of laboratory-grown Arabidopsis thaliana and field-grown cauliflower plants, we show that our approach produces realistic, reliable, and reasonable images of future growth stages. The automatic interpretation of the generated images through neural network-based instance segmentation allows the derivation of various phenotypic traits that describe plant growth.

16.HCRF-Flow: Scene Flow from Point Clouds with Continuous High-order CRFs and Position-aware Flow Embedding ⬇️

Scene flow in 3D point clouds plays an important role in understanding dynamic environments. Although significant advances have been made by deep neural networks, the performance is far from satisfactory as only per-point translational motion is considered, neglecting the constraints of the rigid motion in local regions. To address the issue, we propose to introduce the motion consistency to force the smoothness among neighboring points. In addition, constraints on the rigidity of the local transformation are also added by sharing unique rigid motion parameters for all points within each local region. To this end, a high-order CRFs based relation module (Con-HCRFs) is deployed to explore both point-wise smoothness and region-wise rigidity. To empower the CRFs to have a discriminative unary term, we also introduce a position-aware flow estimation module to be incorporated into the Con-HCRFs. Comprehensive experiments on FlyingThings3D and KITTI show that our proposed framework (HCRF-Flow) achieves state-of-the-art performance and significantly outperforms previous approaches substantially.

17.Cross-Modality Brain Tumor Segmentation via Bidirectional Global-to-Local Unsupervised Domain Adaptation ⬇️

Accurate segmentation of brain tumors from multi-modal Magnetic Resonance (MR) images is essential in brain tumor diagnosis and treatment. However, due to the existence of domain shifts among different modalities, the performance of networks decreases dramatically when training on one modality and performing on another, e.g., train on T1 image while performing on T2 image, which is often required in clinical applications. This also prohibits a network from being trained on labeled data and then transferred to unlabeled data from a different domain. To overcome this, unsupervised domain adaptation (UDA) methods provide effective solutions to alleviate the domain shift between labeled source data and unlabeled target data. In this paper, we propose a novel Bidirectional Global-to-Local (BiGL) adaptation framework under a UDA scheme. Specifically, a bidirectional image synthesis and segmentation module is proposed to segment the brain tumor using the intermediate data distributions generated for the two domains, which includes an image-to-image translator and a shared-weighted segmentation network. Further, a global-to-local consistency learning module is proposed to build robust representation alignments in an integrated way. Extensive experiments on a multi-modal brain MR benchmark dataset demonstrate that the proposed method outperforms several state-of-the-art unsupervised domain adaptation methods by a large margin, while a comprehensive ablation study validates the effectiveness of each key component. The implementation code of our method will be released at \url{this https URL}.

18.EA-Net: Edge-Aware Network for Flow-based Video Frame Interpolation ⬇️

Video frame interpolation can up-convert the frame rate and enhance the video quality. In recent years, although the interpolation performance has achieved great success, image blur usually occurs at the object boundaries owing to the large motion. It has been a long-standing problem, and has not been addressed yet. In this paper, we propose to reduce the image blur and get the clear shape of objects by preserving the edges in the interpolated frames. To this end, the proposed Edge-Aware Network (EA-Net) integrates the edge information into the frame interpolation task. It follows an end-to-end architecture and can be separated into two stages, \emph{i.e.}, edge-guided flow estimation and edge-protected frame synthesis. Specifically, in the flow estimation stage, three edge-aware mechanisms are developed to emphasize the frame edges in estimating flow maps, so that the edge-maps are taken as the auxiliary information to provide more guidance to boost the flow accuracy. In the frame synthesis stage, the flow refinement module is designed to refine the flow map, and the attention module is carried out to adaptively focus on the bidirectional flow maps when synthesizing the intermediate frames. Furthermore, the frame and edge discriminators are adopted to conduct the adversarial training strategy, so as to enhance the reality and clarity of synthesized frames. Experiments on three benchmarks, including Vimeo90k, UCF101 for single-frame interpolation and Adobe240-fps for multi-frame interpolation, have demonstrated the superiority of the proposed EA-Net for the video frame interpolation task.

19.Voxel-level Siamese Representation Learning for Abdominal Multi-Organ Segmentation ⬇️

Recent works in medical image segmentation have actively explored various deep learning architectures or objective functions to encode high-level features from volumetric data owing to limited image annotations. However, most existing approaches tend to ignore cross-volume global context and define context relations in the decision space. In this work, we propose a novel voxel-level Siamese representation learning method for abdominal multi-organ segmentation to improve representation space. The proposed method enforces voxel-wise feature relations in the representation space for leveraging limited datasets more comprehensively to achieve better performance. Inspired by recent progress in contrastive learning, we suppressed voxel-wise relations from the same class to be projected to the same point without using negative samples. Moreover, we introduce a multi-resolution context aggregation method that aggregates features from multiple hidden layers, which encodes both the global and local contexts for segmentation. Our experiments on the multi-organ dataset outperformed the existing approaches by 2% in Dice score coefficient. The qualitative visualizations of the representation spaces demonstrate that the improvements were gained primarily by a disentangled feature space.

20.AudioVisual Video Summarization ⬇️

Audio and vision are two main modalities in video data. Multimodal learning, especially for audiovisual learning, has drawn considerable attention recently, which can boost the performance of various computer vision tasks. However, in video summarization, existing approaches just exploit the visual information while neglect the audio information. In this paper, we argue that the audio modality can assist vision modality to better understand the video content and structure, and further benefit the summarization process. Motivated by this, we propose to jointly exploit the audio and visual information for the video summarization task, and develop an AudioVisual Recurrent Network (AVRN) to achieve this. Specifically, the proposed AVRN can be separated into three parts: 1) the two-stream LSTM is utilized to encode the audio and visual feature sequentially by capturing their temporal dependency. 2) the audiovisual fusion LSTM is employed to fuse the two modalities by exploring the latent consistency between them. 3) the self-attention video encoder is adopted to capture the global dependency in the video. Finally, the fused audiovisual information, and the integrated temporal and global dependencies are jointly used to predict the video summary. Practically, the experimental results on the two benchmarks, \emph{i.e.,} SumMe and TVsum, have demonstrated the effectiveness of each part, and the superiority of AVRN compared to those approaches just exploiting visual information for video summarization.

21.Global Wheat Head Dataset 2021: an update to improve the benchmarking wheat head localization with more diversity ⬇️

The Global Wheat Head Detection (GWHD) dataset was created in 2020 and has assembled 193,634 labelled wheat heads from 4,700 RGB images acquired from various acquisition platforms and 7 countries/institutions. With an associated competition hosted in Kaggle, GWHD has successfully attracted attention from both the computer vision and agricultural science communities. From this first experience in 2020, a few avenues for improvements have been identified, especially from the perspective of data size, head diversity and label reliability. To address these issues, the 2020 dataset has been reexamined, relabeled, and augmented by adding 1,722 images from 5 additional countries, allowing for 81,553 additional wheat heads to be added. We would hence like to release a new version of the Global Wheat Head Detection (GWHD) dataset in 2021, which is bigger, more diverse, and less noisy than the 2020 version. The GWHD 2021 is now publicly available at this http URL and a new data challenge has been organized on AIcrowd to make use of this updated dataset.

22.FGR: Frustum-Aware Geometric Reasoning for Weakly Supervised 3D Vehicle Detection ⬇️

In this paper, we investigate the problem of weakly supervised 3D vehicle detection. Conventional methods for 3D object detection need vast amounts of manually labelled 3D data as supervision signals. However, annotating large datasets requires huge human efforts, especially for 3D area. To tackle this problem, we propose frustum-aware geometric reasoning (FGR) to detect vehicles in point clouds without any 3D annotations. Our method consists of two stages: coarse 3D segmentation and 3D bounding box estimation. For the first stage, a context-aware adaptive region growing algorithm is designed to segment objects based on 2D bounding boxes. Leveraging predicted segmentation masks, we develop an anti-noise approach to estimate 3D bounding boxes in the second stage. Finally 3D pseudo labels generated by our method are utilized to train a 3D detector. Independent of any 3D groundtruth, FGR reaches comparable performance with fully supervised methods on the KITTI dataset. The findings indicate that it is able to accurately detect objects in 3D space with only 2D bounding boxes and sparse point clouds.

23.Leveraging EfficientNet and Contrastive Learning for Accurate Global-scale Location Estimation ⬇️

In this paper, we address the problem of global-scale image geolocation, proposing a mixed classification-retrieval scheme. Unlike other methods that strictly tackle the problem as a classification or retrieval task, we combine the two practices in a unified solution leveraging the advantages of each approach with two different modules. The first leverages the EfficientNet architecture to assign images to a specific geographic cell in a robust way. The second introduces a new residual architecture that is trained with contrastive learning to map input images to an embedding space that minimizes the pairwise geodesic distance of same-location images. For the final location estimation, the two modules are combined with a search-within-cell scheme, where the locations of most similar images from the predicted geographic cell are aggregated based on a spatial clustering scheme. Our approach demonstrates very competitive performance on four public datasets, achieving new state-of-the-art performance in fine granularity scales, i.e., 15.0% at 1km range on Im2GPS3k.

24.Traffic Scenario Clustering by Iterative Optimisation of Self-Supervised Networks Using a Random Forest Activation Pattern Similarity ⬇️

Traffic scenario categorisation is an essential component of automated driving, for e.,g., in motion planning algorithms and their validation. Finding new relevant scenarios without handcrafted steps reduce the required resources for the development of autonomous driving dramatically. In this work, a method is proposed to address this challenge by introducing a clustering technique based on a novel data-adaptive similarity measure, called Random Forest Activation Pattern (RFAP) similarity. The RFAP similarity is generated using a tree encoding scheme in a Random Forest algorithm. The clustering method proposed in this work takes into account that there are labelled scenarios available and the information from the labelled scenarios can help to guide the clustering of unlabelled scenarios. It consists of three steps. First, a self-supervised Convolutional Neural Network~(CNN) is trained on all available traffic scenarios using a defined self-supervised objective. Second, the CNN is fine-tuned for classification of the labelled scenarios. Third, using the labelled and unlabelled scenarios an iterative optimisation procedure is performed for clustering. In the third step at each epoch of the iterative optimisation, the CNN is used as a feature generator for an unsupervised Random Forest. The trained forest, in turn, provides the RFAP similarity to adapt iteratively the feature generation process implemented by the CNN. Extensive experiments and ablation studies have been done on the highD dataset. The proposed method shows superior performance compared to baseline clustering techniques.

25.Class-Incremental Few-Shot Object Detection ⬇️

Conventional detection networks usually need abundant labeled training samples, while humans can learn new concepts incrementally with just a few examples. This paper focuses on a more challenging but realistic class-incremental few-shot object detection problem (iFSD). It aims to incrementally transfer the model for novel objects from only a few annotated samples without catastrophically forgetting the previously learned ones. To tackle this problem, we propose a novel method LEAST, which can transfer with Less forgetting, fEwer training resources, And Stronger Transfer capability. Specifically, we first present the transfer strategy to reduce unnecessary weight adaptation and improve the transfer capability for iFSD. On this basis, we then integrate the knowledge distillation technique using a less resource-consuming approach to alleviate forgetting and propose a novel clustering-based exemplar selection process to preserve more discriminative features previously learned. Being a generic and effective method, LEAST can largely improve the iFSD performance on various benchmarks.

26.Open-set Recognition based on the Combination of Deep Learning and Ensemble Method for Detecting Unknown Traffic Scenarios ⬇️

An understanding and classification of driving scenarios are important for testing and development of autonomous driving functionalities. Machine learning models are useful for scenario classification but most of them assume that data received during the testing are from one of the classes used in the training. This assumption is not true always because of the open environment where vehicles operate. This is addressed by a new machine learning paradigm called open-set recognition. Open-set recognition is the problem of assigning test samples to one of the classes used in training or to an unknown class. This work proposes a combination of Convolutional Neural Networks (CNN) and Random Forest (RF) for open set recognition of traffic scenarios. CNNs are used for the feature generation and the RF algorithm along with extreme value theory for the detection of known and unknown classes. The proposed solution is featured by exploring the vote patterns of trees in RF instead of just majority voting. By inheriting the ensemble nature of RF, the vote pattern of all trees combined with extreme value theory is shown to be well suited for detecting unknown classes. The proposed method has been tested on the highD and OpenTraffic datasets and has demonstrated superior performance in various aspects compared to existing solutions.

27.Shared and Private VAEs with Generative Replay for Continual Learning ⬇️

Continual learning tries to learn new tasks without forgetting previously learned ones. In reality, most of the existing artificial neural network(ANN) models fail, while humans do the same by remembering previous works throughout their life. Although simply storing all past data can alleviate the problem, it needs large memory and often infeasible in real-world applications where last data access is limited. We hypothesize that the model that learns to solve each task continually has some task-specific properties and some task-invariant characteristics. We propose a hybrid continual learning model that is more suitable in real case scenarios to address the issues that has a task-invariant shared variational autoencoder and T task-specific variational autoencoders. Our model combines generative replay and architectural growth to prevent catastrophic forgetting. We show our hybrid model effectively avoids forgetting and achieves state-of-the-art results on visual continual learning benchmarks such as MNIST, Permuted MNIST(QMNIST), CIFAR100, and miniImageNet datasets. We discuss results on a few more datasets, such as SVHN, Fashion-MNIST, EMNIST, and CIFAR10.

28.A Fine-Grained Visual Attention Approach for Fingerspelling Recognition in the Wild ⬇️

Fingerspelling in sign language has been the means of communicating technical terms and proper nouns when they do not have dedicated sign language gestures. Automatic recognition of fingerspelling can help resolve communication barriers when interacting with deaf people. The main challenges prevalent in fingerspelling recognition are the ambiguity in the gestures and strong articulation of the hands. The automatic recognition model should address high inter-class visual similarity and high intra-class variation in the gestures. Most of the existing research in fingerspelling recognition has focused on the dataset collected in a controlled environment. The recent collection of a large-scale annotated fingerspelling dataset in the wild, from social media and online platforms, captures the challenges in a real-world scenario. In this work, we propose a fine-grained visual attention mechanism using the Transformer model for the sequence-to-sequence prediction task in the wild dataset. The fine-grained attention is achieved by utilizing the change in motion of the video frames (optical flow) in sequential context-based attention along with a Transformer encoder model. The unsegmented continuous video dataset is jointly trained by balancing the Connectionist Temporal Classification (CTC) loss and the maximum-entropy loss. The proposed approach can capture better fine-grained attention in a single iteration. Experiment evaluations show that it outperforms the state-of-the-art approaches.

29.Style-Restricted GAN: Multi-Modal Translation with Style Restriction Using Generative Adversarial Networks ⬇️

Unpaired image-to-image translation using Generative Adversarial Networks (GAN) is successful in converting images among multiple domains. Moreover, recent studies have shown a way to diversify the outputs of the generator. However, since there are no restrictions on how the generator diversifies the results, it is likely to translate some unexpected features. In this paper, we propose Style-Restricted GAN (SRGAN), a novel approach to transfer input images into different domains' with different styles, changing the exclusively class-related features. Additionally, instead of KL divergence loss, we adopt 3 new losses to restrict the distribution of the encoded features: batch KL divergence loss, correlation loss, and histogram imitation loss. The study reports quantitative as well as qualitative results with Precision, Recall, Density, and Coverage. The proposed 3 losses lead to the enhancement of the level of diversity compared to the conventional KL loss. In particular, SRGAN is found to be successful in translating with higher diversity and without changing the class-unrelated features in the CelebA face dataset. Our implementation is available at this https URL.

30.Towards Unsupervised Domain Adaptation for Deep Face Recognition under Privacy Constraints via Federated Learning ⬇️

Unsupervised domain adaptation has been widely adopted to generalize models for unlabeled data in a target domain, given labeled data in a source domain, whose data distributions differ from the target domain. However, existing works are inapplicable to face recognition under privacy constraints because they require sharing sensitive face images between two domains. To address this problem, we propose a novel unsupervised federated face recognition approach (FedFR). FedFR improves the performance in the target domain by iteratively aggregating knowledge from the source domain through federated learning. It protects data privacy by transferring models instead of raw data between domains. Besides, we propose a new domain constraint loss (DCL) to regularize source domain training. DCL suppresses the data volume dominance of the source domain. We also enhance a hierarchical clustering algorithm to predict pseudo labels for the unlabeled target domain accurately. To this end, FedFR forms an end-to-end training pipeline: (1) pre-train in the source domain; (2) predict pseudo labels by clustering in the target domain; (3) conduct domain-constrained federated learning across two domains. Extensive experiments and analysis on two newly constructed benchmarks demonstrate the effectiveness of FedFR. It outperforms the baseline and classic methods in the target domain by over 4% on the more realistic benchmark. We believe that FedFR will shed light on applying federated learning to more computer vision tasks under privacy constraints.

31.Differentiable SLAM-net: Learning Particle SLAM for Visual Navigation ⬇️

Simultaneous localization and mapping (SLAM) remains challenging for a number of downstream applications, such as visual robot navigation, because of rapid turns, featureless walls, and poor camera quality. We introduce the Differentiable SLAM Network (SLAM-net) along with a navigation architecture to enable planar robot navigation in previously unseen indoor environments. SLAM-net encodes a particle filter based SLAM algorithm in a differentiable computation graph, and learns task-oriented neural network components by backpropagating through the SLAM algorithm. Because it can optimize all model components jointly for the end-objective, SLAM-net learns to be robust in challenging conditions. We run experiments in the Habitat platform with different real-world RGB and RGB-D datasets. SLAM-net significantly outperforms the widely adapted ORB-SLAM in noisy conditions. Our navigation architecture with SLAM-net improves the state-of-the-art for the Habitat Challenge 2020 PointNav task by a large margin (37% to 64% success). Project website: this http URL

32.Vision Transformers are Robust Learners ⬇️

Transformers, composed of multiple self-attention layers, hold strong promises toward a generic learning primitive applicable to different data modalities, including the recent breakthroughs in computer vision achieving state-of-the-art (SOTA) standard accuracy with better parameter efficiency. Since self-attention helps a model systematically align different components present inside the input data, it leaves grounds to investigate its performance under model robustness benchmarks. In this work, we study the robustness of the Vision Transformer (ViT) against common corruptions and perturbations, distribution shifts, and natural adversarial examples. We use six different diverse ImageNet datasets concerning robust classification to conduct a comprehensive performance comparison of ViT models and SOTA convolutional neural networks (CNNs), Big-Transfer. Through a series of six systematically designed experiments, we then present analyses that provide both quantitative and qualitative indications to explain why ViTs are indeed more robust learners. For example, with fewer parameters and similar dataset and pre-training combinations, ViT gives a top-1 accuracy of 28.10% on ImageNet-A which is 4.3x higher than a comparable variant of BiT. Our analyses on image masking, Fourier spectrum sensitivity, and spread on discrete cosine energy spectrum reveal intriguing properties of ViT attributing to improved robustness. Code for reproducing our experiments is available here: this https URL.

33.Rethinking "Batch" in BatchNorm ⬇️

BatchNorm is a critical building block in modern convolutional neural networks. Its unique property of operating on "batches" instead of individual samples introduces significantly different behaviors from most other operations in deep learning. As a result, it leads to many hidden caveats that can negatively impact model's performance in subtle ways. This paper thoroughly reviews such problems in visual recognition tasks, and shows that a key to address them is to rethink different choices in the concept of "batch" in BatchNorm. By presenting these caveats and their mitigations, we hope this review can help researchers use BatchNorm more effectively.

34.Layerwise Optimization by Gradient Decomposition for Continual Learning ⬇️

Deep neural networks achieve state-of-the-art and sometimes super-human performance across various domains. However, when learning tasks sequentially, the networks easily forget the knowledge of previous tasks, known as "catastrophic forgetting". To achieve the consistencies between the old tasks and the new task, one effective solution is to modify the gradient for update. Previous methods enforce independent gradient constraints for different tasks, while we consider these gradients contain complex information, and propose to leverage inter-task information by gradient decomposition. In particular, the gradient of an old task is decomposed into a part shared by all old tasks and a part specific to that task. The gradient for update should be close to the gradient of the new task, consistent with the gradients shared by all old tasks, and orthogonal to the space spanned by the gradients specific to the old tasks. In this way, our approach encourages common knowledge consolidation without impairing the task-specific knowledge. Furthermore, the optimization is performed for the gradients of each layer separately rather than the concatenation of all gradients as in previous works. This effectively avoids the influence of the magnitude variation of the gradients in different layers. Extensive experiments validate the effectiveness of both gradient-decomposed optimization and layer-wise updates. Our proposed method achieves state-of-the-art results on various benchmarks of continual learning.

35.Prototype-supervised Adversarial Network for Targeted Attack of Deep Hashing ⬇️

Due to its powerful capability of representation learning and high-efficiency computation, deep hashing has made significant progress in large-scale image retrieval. However, deep hashing networks are vulnerable to adversarial examples, which is a practical secure problem but seldom studied in hashing-based retrieval field. In this paper, we propose a novel prototype-supervised adversarial network (ProS-GAN), which formulates a flexible generative architecture for efficient and effective targeted hashing attack. To the best of our knowledge, this is the first generation-based method to attack deep hashing networks. Generally, our proposed framework consists of three parts, i.e., a PrototypeNet, a generator, and a discriminator. Specifically, the designed PrototypeNet embeds the target label into the semantic representation and learns the prototype code as the category-level representative of the target label. Moreover, the semantic representation and the original image are jointly fed into the generator for a flexible targeted attack. Particularly, the prototype code is adopted to supervise the generator to construct the targeted adversarial example by minimizing the Hamming distance between the hash code of the adversarial example and the prototype code. Furthermore, the generator is against the discriminator to simultaneously encourage the adversarial examples visually realistic and the semantic representation informative. Extensive experiments verify that the proposed framework can efficiently produce adversarial examples with better targeted attack performance and transferability over state-of-the-art targeted attack methods of deep hashing. The related codes could be available at this https URL .

36.Algorithmic Principles of Camera-based Respiratory Motion Extraction ⬇️

Measuring the respiratory signal from a video based on body motion has been proposed and recently matured in products for video health monitoring. The core algorithm for this measurement is the estimation of tiny chest/abdominal motions induced by respiration, and the fundamental challenge is motion sensitivity. Though prior arts reported on the validation with real human subjects, there is no thorough/rigorous benchmark to quantify the sensitivities and boundary conditions of motion-based core respiratory algorithms that measure sub-pixel displacement between video frames. In this paper, we designed a setup with a fully-controllable physical phantom to investigate the essence of core algorithms, together with a mathematical model incorporating two motion estimation strategies and three spatial representations, leading to six algorithmic combinations for respiratory signal extraction. Their promises and limitations are discussed and clarified via the phantom benchmark. The insights gained in this paper are intended to improve the understanding and applications of camera-based respiration measurement in health monitoring.

37.Leveraging Semantic Scene Characteristics and Multi-Stream Convolutional Architectures in a Contextual Approach for Video-Based Visual Emotion Recognition in the Wild ⬇️

In this work we tackle the task of video-based visual emotion recognition in the wild. Standard methodologies that rely solely on the extraction of bodily and facial features often fall short of accurate emotion prediction in cases where the aforementioned sources of affective information are inaccessible due to head/body orientation, low resolution and poor illumination. We aspire to alleviate this problem by leveraging visual context in the form of scene characteristics and attributes, as part of a broader emotion recognition framework. Temporal Segment Networks (TSN) constitute the backbone of our proposed model. Apart from the RGB input modality, we make use of dense Optical Flow, following an intuitive multi-stream approach for a more effective encoding of motion. Furthermore, we shift our attention towards skeleton-based learning and leverage action-centric data as means of pre-training a Spatial-Temporal Graph Convolutional Network (ST-GCN) for the task of emotion recognition. Our extensive experiments on the challenging Body Language Dataset (BoLD) verify the superiority of our methods over existing approaches, while by properly incorporating all of the aforementioned modules in a network ensemble, we manage to surpass the previous best published recognition scores, by a large margin.

38.Fast-GANFIT: Generative Adversarial Network for High Fidelity 3D Face Reconstruction ⬇️

A lot of work has been done towards reconstructing the 3D facial structure from single images by capitalizing on the power of Deep Convolutional Neural Networks (DCNNs). In the recent works, the texture features either correspond to components of a linear texture space or are learned by auto-encoders directly from in-the-wild images. In all cases, the quality of the facial texture reconstruction is still not capable of modeling facial texture with high-frequency details. In this paper, we take a radically different approach and harness the power of Generative Adversarial Networks (GANs) and DCNNs in order to reconstruct the facial texture and shape from single images. That is, we utilize GANs to train a very powerful facial texture prior \edit{from a large-scale 3D texture dataset}. Then, we revisit the original 3D Morphable Models (3DMMs) fitting making use of non-linear optimization to find the optimal latent parameters that best reconstruct the test image but under a new perspective. In order to be robust towards initialisation and expedite the fitting process, we propose a novel self-supervised regression based approach. We demonstrate excellent results in photorealistic and identity preserving 3D face reconstructions and achieve for the first time, to the best of our knowledge, facial texture reconstruction with high-frequency details.

39.Uncertainty in Minimum Cost Multicuts for Image and Motion Segmentation ⬇️

The minimum cost lifted multicut approach has proven practically good performance in a wide range of applications such as image decomposition, mesh segmentation, multiple object tracking, and motion segmentation. It addresses such problems in a graph-based model, where real-valued costs are assigned to the edges between entities such that the minimum cut decomposes the graph into an optimal number of segments. Driven by a probabilistic formulation of minimum cost multicuts, we provide a measure for the uncertainties of the decisions made during the optimization. We argue that access to such uncertainties is crucial for many practical applications and conduct an evaluation by means of sparsifications on three different, widely used datasets in the context of image decomposition (BSDS-500) and motion segmentation (DAVIS2016 and FBMS59) in terms of variation of information (VI) and Rand index (RI).

40.TSDF++: A Multi-Object Formulation for Dynamic Object Tracking and Reconstruction ⬇️

The ability to simultaneously track and reconstruct multiple objects moving in the scene is of the utmost importance for robotic tasks such as autonomous navigation and interaction. Virtually all of the previous attempts to map multiple dynamic objects have evolved to store individual objects in separate reconstruction volumes and track the relative pose between them. While simple and intuitive, such formulation does not scale well with respect to the number of objects in the scene and introduces the need for an explicit occlusion handling strategy. In contrast, we propose a map representation that allows maintaining a single volume for the entire scene and all the objects therein. To this end, we introduce a novel multi-object TSDF formulation that can encode multiple object surfaces at any given location in the map. In a multiple dynamic object tracking and reconstruction scenario, our representation allows maintaining accurate reconstruction of surfaces even while they become temporarily occluded by other objects moving in their proximity. We evaluate the proposed TSDF++ formulation on a public synthetic dataset and demonstrate its ability to preserve reconstructions of occluded surfaces when compared to the standard TSDF map representation.

41.3D to 4D Facial Expressions Generation Guided by Landmarks ⬇️

While deep learning-based 3D face generation has made a progress recently, the problem of dynamic 3D (4D) facial expression synthesis is less investigated. In this paper, we propose a novel solution to the following question: given one input 3D neutral face, can we generate dynamic 3D (4D) facial expressions from it? To tackle this problem, we first propose a mesh encoder-decoder architecture (Expr-ED) that exploits a set of 3D landmarks to generate an expressive 3D face from its neutral counterpart. Then, we extend it to 4D by modeling the temporal dynamics of facial expressions using a manifold-valued GAN capable of generating a sequence of 3D landmarks from an expression label (Motion3DGAN). The generated landmarks are fed into the mesh encoder-decoder, ultimately producing a sequence of 3D expressive faces. By decoupling the two steps, we separately address the non-linearity induced by the mesh deformation and motion dynamics. The experimental results on the CoMA dataset show that our mesh encoder-decoder guided by landmarks brings a significant improvement with respect to other landmark-based 3D fitting approaches, and that we can generate high quality dynamic facial expressions. This framework further enables the 3D expression intensity to be continuously adapted from low to high intensity. Finally, we show our framework can be applied to other tasks, such as 2D-3D facial expression transfer.

42.MultiSports: A Multi-Person Video Dataset of Spatio-Temporally Localized Sports Actions ⬇️

Spatio-temporal action detection is an important and challenging problem in video understanding. The existing action detection benchmarks are limited in aspects of small numbers of instances in a trimmed video or relatively low-level atomic actions. This paper aims to present a new multi-person dataset of spatio-temporal localized sports actions, coined as MultiSports. We first analyze the important ingredients of constructing a realistic and challenging dataset for spatio-temporal action detection by proposing three criteria: (1) motion dependent identification, (2) with well-defined boundaries, (3) relatively high-level classes. Based on these guidelines, we build the dataset of Multi-Sports v1.0 by selecting 4 sports classes, collecting around 3200 video clips, and annotating around 37790 action instances with 907k bounding boxes. Our datasets are characterized with important properties of strong diversity, detailed annotation, and high quality. Our MultiSports, with its realistic setting and dense annotations, exposes the intrinsic challenge of action localization. To benchmark this, we adapt several representative methods to our dataset and give an in-depth analysis on the difficulty of action localization in our dataset. We hope our MultiSports can serve as a standard benchmark for spatio-temporal action detection in the future. Our dataset website is at this https URL.

43.Is Image Size Important? A Robustness Comparison of Deep Learning Methods for Multi-scale Cell Image Classification Tasks: from Convolutional Neural Networks to Visual Transformers ⬇️

Cervical cancer is a very common and fatal cancer in women, but it can be prevented through early examination and treatment. Cytopathology images are often used to screen for cancer. Then, because of the possibility of artificial errors due to the large number of this method, the computer-aided diagnosis system based on deep learning is developed. The image input required by the deep learning method is usually consistent, but the size of the clinical medical image is inconsistent. The internal information is lost after resizing the image directly, so it is unreasonable. A lot of research is to directly resize the image, and the results are still robust. In order to find a reasonable explanation, 22 deep learning models are used to process images of different scales, and experiments are conducted on the SIPaKMeD dataset. The conclusion is that the deep learning method is very robust to the size changes of images. This conclusion is also validated on the Herlev dataset.

44.Survey of Visual-Semantic Embedding Methods for Zero-Shot Image Retrieval ⬇️

Visual-semantic embedding is an interesting research topic because it is useful for various tasks, such as visual question answering (VQA), image-text retrieval, image captioning, and scene graph generation. In this paper, we focus on zero-shot image retrieval using sentences as queries and present a survey of the technological trends in this area. First, we provide a comprehensive overview of the history of the technology, starting with a discussion of the early studies of image-to-text matching and how the technology has evolved over time. In addition, a description of the datasets commonly used in experiments and a comparison of the evaluation results of each method are presented. We also introduce the implementation available on github for use in confirming the accuracy of experiments and for further improvement. We hope that this survey paper will encourage researchers to further develop their research on bridging images and languages.

45.Semi-supervised Contrastive Learning with Similarity Co-calibration ⬇️

Semi-supervised learning acts as an effective way to leverage massive unlabeled data. In this paper, we propose a novel training strategy, termed as Semi-supervised Contrastive Learning (SsCL), which combines the well-known contrastive loss in self-supervised learning with the cross entropy loss in semi-supervised learning, and jointly optimizes the two objectives in an end-to-end way. The highlight is that different from self-training based semi-supervised learning that conducts prediction and retraining over the same model weights, SsCL interchanges the predictions over the unlabeled data between the two branches, and thus formulates a co-calibration procedure, which we find is beneficial for better prediction and avoid being trapped in local minimum. Towards this goal, the contrastive loss branch models pairwise similarities among samples, using the nearest neighborhood generated from the cross entropy branch, and in turn calibrates the prediction distribution of the cross entropy branch with the contrastive similarity. We show that SsCL produces more discriminative representation and is beneficial to few shot learning. Notably, on ImageNet with ResNet50 as the backbone, SsCL achieves 60.2% and 72.1% top-1 accuracy with 1% and 10% labeled samples, respectively, which significantly outperforms the baseline, and is better than previous semi-supervised and self-supervised methods.

46.BDANet: Multiscale Convolutional Neural Network with Cross-directional Attention for Building Damage Assessment from Satellite Images ⬇️

Fast and effective responses are required when a natural disaster (e.g., earthquake, hurricane, etc.) strikes. Building damage assessment from satellite imagery is critical before relief effort is deployed. With a pair of pre- and post-disaster satellite images, building damage assessment aims at predicting the extent of damage to buildings. With the powerful ability of feature representation, deep neural networks have been successfully applied to building damage assessment. Most existing works simply concatenate pre- and post-disaster images as input of a deep neural network without considering their correlations. In this paper, we propose a novel two-stage convolutional neural network for Building Damage Assessment, called BDANet. In the first stage, a U-Net is used to extract the locations of buildings. Then the network weights from the first stage are shared in the second stage for building damage assessment. In the second stage, a two-branch multi-scale U-Net is employed as backbone, where pre- and post-disaster images are fed into the network separately. A cross-directional attention module is proposed to explore the correlations between pre- and post-disaster images. Moreover, CutMix data augmentation is exploited to tackle the challenge of difficult classes. The proposed method achieves state-of-the-art performance on a large-scale dataset -- xBD. The code is available at this https URL.

47.ExSinGAN: Learning an Explainable Generative Model from a Single Image ⬇️

Generating images from a single sample, as a newly developing branch of image synthesis, has attracted extensive attention. In this paper, we formulate this problem as sampling from the conditional distribution of a single image, and propose a hierarchical framework that simplifies the learning of the intricate conditional distributions through the successive learning of the distributions about structure, semantics and texture, making the process of learning and generation comprehensible. On this basis, we design ExSinGAN composed of three cascaded GANs for learning an explainable generative model from a given image, where the cascaded GANs model the distributions about structure, semantics and texture successively. ExSinGAN is learned not only from the internal patches of the given image as the previous works did, but also from the external prior obtained by the GAN inversion technique. Benefiting from the appropriate combination of internal and external information, ExSinGAN has a more powerful capability of generation and competitive generalization ability for the image manipulation tasks compared with prior works.

48.Neighbourhood-guided Feature Reconstruction for Occluded Person Re-Identification ⬇️

Person images captured by surveillance cameras are often occluded by various obstacles, which lead to defective feature representation and harm person re-identification (Re-ID) performance. To tackle this challenge, we propose to reconstruct the feature representation of occluded parts by fully exploiting the information of its neighborhood in a gallery image set. Specifically, we first introduce a visible part-based feature by body mask for each person image. Then we identify its neighboring samples using the visible features and reconstruct the representation of the full body by an outlier-removable graph neural network with all the neighboring samples as input. Extensive experiments show that the proposed approach obtains significant improvements. In the large-scale Occluded-DukeMTMC benchmark, our approach achieves 64.2% mAP and 67.6% rank-1 accuracy which outperforms the state-of-the-art approaches by large margins, i.e.,20.4% and 12.5%, respectively, indicating the effectiveness of our method on occluded Re-ID problem.

49.Unsupervised Super-Resolution of Satellite Imagery for High Fidelity Material Label Transfer ⬇️

Urban material recognition in remote sensing imagery is a highly relevant, yet extremely challenging problem due to the difficulty of obtaining human annotations, especially on low resolution satellite images. To this end, we propose an unsupervised domain adaptation based approach using adversarial learning. We aim to harvest information from smaller quantities of high resolution data (source domain) and utilize the same to super-resolve low resolution imagery (target domain). This can potentially aid in semantic as well as material label transfer from a richly annotated source to a target domain.

50.Mask-Guided Discovery of Semantic Manifolds in Generative Models ⬇️

Advances in the realm of Generative Adversarial Networks (GANs) have led to architectures capable of producing amazingly realistic images such as StyleGAN2, which, when trained on the FFHQ dataset, generates images of human faces from random vectors in a lower-dimensional latent space. Unfortunately, this space is entangled - translating a latent vector along its axes does not correspond to a meaningful transformation in the output space (e.g., smiling mouth, squinting eyes). The model behaves as a black box, providing neither control over its output nor insight into the structures it has learned from the data. We present a method to explore the manifolds of changes of spatially localized regions of the face. Our method discovers smoothly varying sequences of latent vectors along these manifolds suitable for creating animations. Unlike existing disentanglement methods that either require labelled data or explicitly alter internal model parameters, our method is an optimization-based approach guided by a custom loss function and manually defined region of change. Our code is open-sourced, which can be found, along with supplementary results, on our project page: this https URL

51.Mean Shift for Self-Supervised Learning ⬇️

Most recent self-supervised learning (SSL) algorithms learn features by contrasting between instances of images or by clustering the images and then contrasting between the image clusters. We introduce a simple mean-shift algorithm that learns representations by grouping images together without contrasting between them or adopting much of prior on the structure of the clusters. We simply "shift" the embedding of each image to be close to the "mean" of its neighbors. Since in our setting, the closest neighbor is always another augmentation of the same image, our model will be identical to BYOL when using only one nearest neighbor instead of 5 as used in our experiments. Our model achieves 72.4% on ImageNet linear evaluation with ResNet50 at 200 epochs outperforming BYOL. Our code is available here: this https URL

52.Composite Localization for Human Pose Estimation ⬇️

The existing human pose estimation methods are confronted with inaccurate long-distance regression or high computational cost due to the complex learning objectives. This work proposes a novel deep learning framework for human pose estimation called composite localization to divide the complex learning objective into two simpler ones: a sparse heatmap to find the keypoint's approximate location and two short-distance offsetmaps to obtain its final precise coordinates. To realize the framework, we construct two types of composite localization networks: CLNet-ResNet and CLNet-Hourglass. We evaluate the networks on three benchmark datasets, including the Leeds Sports Pose dataset, the MPII Human Pose dataset, and the COCO keypoints detection dataset. The experimental results show that our CLNet-ResNet50 outperforms SimpleBaseline by 1.14% with about 1/2 GFLOPs. Our CLNet-Hourglass outperforms the original stacked-hourglass by 4.45% on COCO.

53.AgeFlow: Conditional Age Progression and Regression with Normalizing Flows ⬇️

Age progression and regression aim to synthesize photorealistic appearance of a given face image with aging and rejuvenation effects, respectively. Existing generative adversarial networks (GANs) based methods suffer from the following three major issues: 1) unstable training introducing strong ghost artifacts in the generated faces, 2) unpaired training leading to unexpected changes in facial attributes such as genders and races, and 3) non-bijective age mappings increasing the uncertainty in the face transformation. To overcome these issues, this paper proposes a novel framework, termed AgeFlow, to integrate the advantages of both flow-based models and GANs. The proposed AgeFlow contains three parts: an encoder that maps a given face to a latent space through an invertible neural network, a novel invertible conditional translation module (ICTM) that translates the source latent vector to target one, and a decoder that reconstructs the generated face from the target latent vector using the same encoder network; all parts are invertible achieving bijective age mappings. The novelties of ICTM are two-fold. First, we propose an attribute-aware knowledge distillation to learn the manipulation direction of age progression while keeping other unrelated attributes unchanged, alleviating unexpected changes in facial attributes. Second, we propose to use GANs in the latent space to ensure the learned latent vector indistinguishable from the real ones, which is much easier than traditional use of GANs in the image domain. Experimental results demonstrate superior performance over existing GANs-based methods on two benchmarked datasets. The source code is available at this https URL.

54.Brain Inspired Object Recognition System ⬇️

This paper presents a new proposal of an efficient computational model of face and object recognition which uses cues from the distributed face and object recognition mechanism of the brain, and by gathering engineering equivalent of these cues from existing literature. Three distinct and widely used features, Histogram of Oriented Gradients, Local Binary Patterns, and Principal components extracted from target images are used in a manner which is simple, and yet effective. Our model uses multi-layer perceptrons (MLP) to classify these three features and fuse them at the decision level using sum rule. A computational theory is first developed by using concepts from the information processing mechanism of the brain. Extensive experiments are carried out using fifteen publicly available datasets to validate the performance of our proposed model in recognizing faces and objects with extreme variation of illumination, pose angle, expression, and background. Results obtained are extremely promising when compared with other face and object recognition algorithms including CNN and deep learning based methods. This highlights that simple computational processes, if clubbed properly, can produce competing performance with best algorithms.

55.Aerial-PASS: Panoramic Annular Scene Segmentation in Drone Videos ⬇️

Aerial pixel-wise scene perception of the surrounding environment is an important task for UAVs (Unmanned Aerial Vehicles). Previous research works mainly adopt conventional pinhole cameras or fisheye cameras as the imaging device. However, these imaging systems cannot achieve large Field of View (FoV), small size, and lightweight at the same time. To this end, we design a UAV system with a Panoramic Annular Lens (PAL), which has the characteristics of small size, low weight, and a 360-degree annular FoV. A lightweight panoramic annular semantic segmentation neural network model is designed to achieve high-accuracy and real-time scene parsing. In addition, we present the first drone-perspective panoramic scene segmentation dataset Aerial-PASS, with annotated labels of track, field, and others. A comprehensive variety of experiments shows that the designed system performs satisfactorily in aerial panoramic scene parsing. In particular, our proposed model strikes an excellent trade-off between segmentation performance and inference speed suitable, validated on both public street-scene and our established aerial-scene datasets.

56.Are Convolutional Neural Networks or Transformers more like human vision? ⬇️

Modern machine learning models for computer vision exceed humans in accuracy on specific visual recognition tasks, notably on datasets like ImageNet. However, high accuracy can be achieved in many ways. The particular decision function found by a machine learning system is determined not only by the data to which the system is exposed, but also the inductive biases of the model, which are typically harder to characterize. In this work, we follow a recent trend of in-depth behavioral analyses of neural network models that go beyond accuracy as an evaluation metric by looking at patterns of errors. Our focus is on comparing a suite of standard Convolutional Neural Networks (CNNs) and a recently-proposed attention-based network, the Vision Transformer (ViT), which relaxes the translation-invariance constraint of CNNs and therefore represents a model with a weaker set of inductive biases. Attention-based networks have previously been shown to achieve higher accuracy than CNNs on vision tasks, and we demonstrate, using new metrics for examining error consistency with more granularity, that their errors are also more consistent with those of humans. These results have implications both for building more human-like vision models, as well as for understanding visual object recognition in humans.

57.Cross-Modal Progressive Comprehension for Referring Segmentation ⬇️

Given a natural language expression and an image/video, the goal of referring segmentation is to produce the pixel-level masks of the entities described by the subject of the expression. Previous approaches tackle this problem by implicit feature interaction and fusion between visual and linguistic modalities in a one-stage manner. However, human tends to solve the referring problem in a progressive manner based on informative words in the expression, i.e., first roughly locating candidate entities and then distinguishing the target one. In this paper, we propose a Cross-Modal Progressive Comprehension (CMPC) scheme to effectively mimic human behaviors and implement it as a CMPC-I (Image) module and a CMPC-V (Video) module to improve referring image and video segmentation models. For image data, our CMPC-I module first employs entity and attribute words to perceive all the related entities that might be considered by the expression. Then, the relational words are adopted to highlight the target entity as well as suppress other irrelevant ones by spatial graph reasoning. For video data, our CMPC-V module further exploits action words based on CMPC-I to highlight the correct entity matched with the action cues by temporal graph reasoning. In addition to the CMPC, we also introduce a simple yet effective Text-Guided Feature Exchange (TGFE) module to integrate the reasoned multimodal features corresponding to different levels in the visual backbone under the guidance of textual information. In this way, multi-level features can communicate with each other and be mutually refined based on the textual context. Combining CMPC-I or CMPC-V with TGFE can form our image or video version referring segmentation frameworks and our frameworks achieve new state-of-the-art performances on four referring image segmentation benchmarks and three referring video segmentation benchmarks respectively.

58.Stacked Deep Multi-Scale Hierarchical Network for Fast Bokeh Effect Rendering from a Single Image ⬇️

The Bokeh Effect is one of the most desirable effects in photography for rendering artistic and aesthetic photos. Usually, it requires a DSLR camera with different aperture and shutter settings and certain photography skills to generate this effect. In smartphones, computational methods and additional sensors are used to overcome the physical lens and sensor limitations to achieve such effect. Most of the existing methods utilized additional sensor's data or pretrained network for fine depth estimation of the scene and sometimes use portrait segmentation pretrained network module to segment salient objects in the image. Because of these reasons, networks have many parameters, become runtime intensive and unable to run in mid-range devices. In this paper, we used an end-to-end Deep Multi-Scale Hierarchical Network (DMSHN) model for direct Bokeh effect rendering of images captured from the monocular camera. To further improve the perceptual quality of such effect, a stacked model consisting of two DMSHN modules is also proposed. Our model does not rely on any pretrained network module for Monocular Depth Estimation or Saliency Detection, thus significantly reducing the size of model and run time. Stacked DMSHN achieves state-of-the-art results on a large scale EBB! dataset with around 6x less runtime compared to the current state-of-the-art model in processing HD quality images.

59.FloorPlanCAD: A Large-Scale CAD Drawing Dataset for Panoptic Symbol Spotting ⬇️

Access to large and diverse computer-aided design (CAD) drawings is critical for developing symbol spotting algorithms. In this paper, we present FloorPlanCAD, a large-scale real-world CAD drawing dataset containing over 10,000 floor plans, ranging from residential to commercial buildings. CAD drawings in the dataset are all represented as vector graphics, which enable us to provide line-grained annotations of 30 object categories. Equipped by such annotations, we introduce the task of panoptic symbol spotting, which requires to spot not only instances of countable things, but also the semantic of uncountable stuff. Aiming to solve this task, we propose a novel method by combining Graph Convolutional Networks (GCNs) with Convolutional Neural Networks (CNNs), which captures both non-Euclidean and Euclidean features and can be trained end-to-end. The proposed CNN-GCN method achieved state-of-the-art (SOTA) performance on the task of semantic symbol spotting, and help us build a baseline network for the panoptic symbol spotting task. Our contributions are three-fold: 1) to the best of our knowledge, the presented CAD drawing dataset is the first of its kind; 2) the panoptic symbol spotting task considers the spotting of both thing instances and stuff semantic as one recognition problem; and 3) we presented a baseline solution to the panoptic symbol spotting task based on a novel CNN-GCN method, which achieved SOTA performance on semantic symbol spotting. We believe that these contributions will boost research in related areas.

60.One for All: An End-to-End Compact Solution for Hand Gesture Recognition ⬇️

The HGR is a quite challenging task as its performance is influenced by various aspects such as illumination variations, cluttered backgrounds, spontaneous capture, etc. The conventional CNN networks for HGR are following two stage pipeline to deal with the various challenges: complex signs, illumination variations, complex and cluttered backgrounds. The existing approaches needs expert expertise as well as auxiliary computation at stage 1 to remove the complexities from the input images. Therefore, in this paper, we proposes an novel end-to-end compact CNN framework: fine grained feature attentive network for hand gesture recognition (Fit-Hand) to solve the challenges as discussed above. The pipeline of the proposed architecture consists of two main units: FineFeat module and dilated convolutional (Conv) layer. The FineFeat module extracts fine grained feature maps by employing attention mechanism over multiscale receptive fields. The attention mechanism is introduced to capture effective features by enlarging the average behaviour of multi-scale responses. Moreover, dilated convolution provides global features of hand gestures through a larger receptive field. In addition, integrated layer is also utilized to combine the features of FineFeat module and dilated layer which enhances the discriminability of the network by capturing complementary context information of hand postures. The effectiveness of Fit- Hand is evaluated by using subject dependent (SD) and subject independent (SI) validation setup over seven benchmark datasets: MUGD-I, MUGD-II, MUGD-III, MUGD-IV, MUGD-V, Finger Spelling and OUHANDS, respectively. Furthermore, to investigate the deep insights of the proposed Fit-Hand framework, we performed ten ablation study.

61.Move2Hear: Active Audio-Visual Source Separation ⬇️

We introduce the active audio-visual source separation problem, where an agent must move intelligently in order to better isolate the sounds coming from an object of interest in its environment. The agent hears multiple audio sources simultaneously (e.g., a person speaking down the hall in a noisy household) and must use its eyes and ears to automatically separate out the sounds originating from the target object within a limited time budget. Towards this goal, we introduce a reinforcement learning approach that trains movement policies controlling the agent's camera and microphone placement over time, guided by the improvement in predicted audio separation quality. We demonstrate our approach in scenarios motivated by both augmented reality (system is already co-located with the target object) and mobile robotics (agent begins arbitrarily far from the target object). Using state-of-the-art realistic audio-visual simulations in 3D environments, we demonstrate our model's ability to find minimal movement sequences with maximal payoff for audio source separation. Project: this http URL.

62.Show Why the Answer is Correct! Towards Explainable AI using Compositional Temporal Attention ⬇️

Visual Question Answering (VQA) models have achieved significant success in recent times. Despite the success of VQA models, they are mostly black-box models providing no reasoning about the predicted answer, thus raising questions for their applicability in safety-critical such as autonomous systems and cyber-security. Current state of the art fail to better complex questions and thus are unable to exploit compositionality. To minimize the black-box effect of these models and also to make them better exploit compositionality, we propose a Dynamic Neural Network (DMN), which can understand a particular question and then dynamically assemble various relatively shallow deep learning modules from a pool of modules to form a network. We incorporate compositional temporal attention to these deep learning based modules to increase compositionality exploitation. This results in achieving better understanding of complex questions and also provides reasoning as to why the module predicts a particular answer. Experimental analysis on the two benchmark datasets, VQA2.0 and CLEVR, depicts that our model outperforms the previous approaches for Visual Question Answering task as well as provides better reasoning, thus making it reliable for mission critical applications like safety and security.

63.Regularized Deep Linear Discriminant Analysis ⬇️

As a non-linear extension of the classic Linear Discriminant Analysis(LDA), Deep Linear Discriminant Analysis(DLDA) replaces the original Categorical Cross Entropy(CCE) loss function with eigenvalue-based loss function to make a deep neural network(DNN) able to learn linearly separable hidden representations. In this paper, we first point out DLDA focuses on training the cooperative discriminative ability of all the dimensions in the latent subspace, while put less emphasis on training the separable capacity of single dimension. To improve DLDA, a regularization method on within-class scatter matrix is proposed to strengthen the discriminative ability of each dimension, and also keep them complement each other. Experiment results on STL-10, CIFAR-10 and Pediatric Pneumonic Chest X-ray Dataset showed that our proposed regularization method Regularized Deep Linear Discriminant Analysis(RDLDA) outperformed DLDA and conventional neural network with CCE as objective. To further improve the discriminative ability of RDLDA in the local space, an algorithm named Subclass RDLDA is also proposed.

64.FDDH: Fast Discriminative Discrete Hashing for Large-Scale Cross-Modal Retrieval ⬇️

Cross-modal hashing, favored for its effectiveness and efficiency, has received wide attention to facilitating efficient retrieval across different modalities. Nevertheless, most existing methods do not sufficiently exploit the discriminative power of semantic information when learning the hash codes, while often involving time-consuming training procedure for handling the large-scale dataset. To tackle these issues, we formulate the learning of similarity-preserving hash codes in terms of orthogonally rotating the semantic data so as to minimize the quantization loss of mapping such data to hamming space, and propose an efficient Fast Discriminative Discrete Hashing (FDDH) approach for large-scale cross-modal retrieval. More specifically, FDDH introduces an orthogonal basis to regress the targeted hash codes of training examples to their corresponding semantic labels, and utilizes "-dragging technique to provide provable large semantic margins. Accordingly, the discriminative power of semantic information can be explicitly captured and maximized. Moreover, an orthogonal transformation scheme is further proposed to map the nonlinear embedding data into the semantic subspace, which can well guarantee the semantic consistency between the data feature and its semantic representation. Consequently, an efficient closed form solution is derived for discriminative hash code learning, which is very computationally efficient. In addition, an effective and stable online learning strategy is presented for optimizing modality-specific projection functions, featuring adaptivity to different training sizes and streaming data. The proposed FDDH approach theoretically approximates the bi-Lipschitz continuity, runs sufficiently fast, and also significantly improves the retrieval performance over the state-of-the-art methods. The source code is released at: this https URL.

65.Can self-training identify suspicious ugly duckling lesions? ⬇️

One commonly used clinical approach towards detecting melanomas recognises the existence of Ugly Duckling nevi, or skin lesions which look different from the other lesions on the same patient. An automatic method of detecting and analysing these lesions would help to standardize studies, compared with manual screening methods. However, it is difficult to obtain expertly-labelled images for ugly duckling lesions. We therefore propose to use self-supervised machine learning to automatically detect outlier lesions. We first automatically detect and extract all the lesions from a wide-field skin image, and calculate an embedding for each detected lesion in a patient image, based on automatically identified features. These embeddings are then used to calculate the L2 distances as a way to measure dissimilarity. Using this deep learning method, Ugly Ducklings are identified as outliers which should deserve more attention from the examining physician. We evaluate through comparison with dermatologists, and achieve a sensitivity rate of 72.1% and diagnostic accuracy of 94.2% on the held-out test set.

66.A Large Visual, Qualitative and Quantitative Dataset of Web Pages ⬇️

The World Wide Web is not only one of the most important platforms of communication and information at present, but also an area of growing interest for scientific research. This motivates a lot of work and projects that require large amounts of data. However, there is no dataset that integrates the parameters and visual appearance of Web pages, because its collection is a costly task in terms of time and effort. With the support of various computer tools and programming scripts, we have created a large dataset of 49,438 Web pages. It consists of visual, textual and numerical data types, includes all countries worldwide, and considers a broad range of topics such as art, entertainment, economy, business, education, government, news, media, science, and environment, covering different cultural characteristics and varied design preferences. In this paper, we describe the process of collecting, debugging and publishing the final product, which is freely available. To demonstrate the usefulness of our dataset, we expose a binary classification model for detecting error Web pages, and a multi-class Web subject-based categorization, both problems using convolutional neural networks.

67.NeLF: Practical Novel View Synthesis with Neural Light Field ⬇️

In this paper, we present a practical and robust deep learning solution for the novel view synthesis of complex scenes. In our approach, a continuous scene is represented as a light field, i.e., a set of rays, each of which has a corresponding color. We adopt a 4D parameterization of the light field. We then formulate the light field as a 4D function that maps 4D coordinates to corresponding color values. We train a deep fully connected network to optimize this function. Then, the scene-specific model is used to synthesize novel views. Previous light field approaches usually require dense view sampling to reliably render high-quality novel views. Our method can render novel views by sampling rays and querying the color for each ray from the network directly; thus enabling fast light field rendering with a very sparse set of input images. Our method achieves state-of-the-art novel view synthesis results while maintaining an interactive frame rate.

68.MutualNet: Adaptive ConvNet via Mutual Learning from Different Model Configurations ⬇️

Most existing deep neural networks are static, which means they can only do inference at a fixed complexity. But the resource budget can vary substantially across different devices. Even on a single device, the affordable budget can change with different scenarios, and repeatedly training networks for each required budget would be incredibly expensive. Therefore, in this work, we propose a general method called MutualNet to train a single network that can run at a diverse set of resource constraints. Our method trains a cohort of model configurations with various network widths and input resolutions. This mutual learning scheme not only allows the model to run at different width-resolution configurations but also transfers the unique knowledge among these configurations, helping the model to learn stronger representations overall. MutualNet is a general training methodology that can be applied to various network structures (e.g., 2D networks: MobileNets, ResNet, 3D networks: SlowFast, X3D) and various tasks (e.g., image classification, object detection, segmentation, and action recognition), and is demonstrated to achieve consistent improvements on a variety of datasets. Since we only train the model once, it also greatly reduces the training cost compared to independently training several models. Surprisingly, MutualNet can also be used to significantly boost the performance of a single network, if dynamic resource constraint is not a concern. In summary, MutualNet is a unified method for both static and adaptive, 2D and 3D networks. Codes and pre-trained models are available at \url{this https URL}.

69.Momentum Contrastive Voxel-wise Representation Learning for Semi-supervised Volumetric Medical Image Segmentation ⬇️

Automated segmentation in medical image analysis is a challenging task that requires a large amount of manually labeled data. However, manually annotating medical data is often laborious, and most existing learning-based approaches fail to accurately delineate object boundaries without effective geometric constraints. Contrastive learning, a sub-area of self-supervised learning, has recently been noted as a promising direction in multiple application fields. In this work, we present a novel Contrastive Voxel-wise Representation Learning (CVRL) method with geometric constraints to learn global-local visual representations for volumetric medical image segmentation with limited annotations. Our framework can effectively learn global and local features by capturing 3D spatial context and rich anatomical information. Specifically, we introduce a voxel-to-volume contrastive algorithm to learn global information from 3D images, and propose to perform local voxel-to-voxel contrast to explicitly make use of local cues in the embedding space. Moreover, we integrate an elastic interaction-based active contour model as a geometric regularization term to enable fast and reliable object delineations in an end-to-end learning manner. Results on the Atrial Segmentation Challenge dataset demonstrate superiority of our proposed scheme, especially in a setting with a very limited number of annotated data.

70.Face Attributes as Cues for Deep Face Recognition Understanding ⬇️

Deeply learned representations are the state-of-the-art descriptors for face recognition methods. These representations encode latent features that are difficult to explain, compromising the confidence and interpretability of their predictions. Most attempts to explain deep features are visualization techniques that are often open to interpretation. Instead of relying only on visualizations, we use the outputs of hidden layers to predict face attributes. The obtained performance is an indicator of how well the attribute is implicitly learned in that layer of the network. Using a variable selection technique, we also analyze how these semantic concepts are distributed inside each layer, establishing the precise location of relevant neurons for each attribute. According to our experiments, gender, eyeglasses and hat usage can be predicted with over 96% accuracy even when only a single neural output is used to predict each attribute. These performances are less than 3 percentage points lower than the ones achieved by deep supervised face attribute networks. In summary, our experiments show that, inside DCNNs optimized for face identification, there exists latent neurons encoding face attributes almost as accurately as DCNNs optimized for these attributes.

71.SMURF: Self-Teaching Multi-Frame Unsupervised RAFT with Full-Image Warping ⬇️

We present SMURF, a method for unsupervised learning of optical flow that improves state of the art on all benchmarks by $36%$ to $40%$ (over the prior best method UFlow) and even outperforms several supervised approaches such as PWC-Net and FlowNet2. Our method integrates architecture improvements from supervised optical flow, i.e. the RAFT model, with new ideas for unsupervised learning that include a sequence-aware self-supervision loss, a technique for handling out-of-frame motion, and an approach for learning effectively from multi-frame video data while still only requiring two frames for inference.

72.Pay Attention to MLPs ⬇️

Transformers have become one of the most important architectural innovations in deep learning and have enabled many breakthroughs over the past few years. Here we propose a simple attention-free network architecture, gMLP, based solely on MLPs with gating, and show that it can perform as well as Transformers in key language and vision applications. Our comparisons show that self-attention is not critical for Vision Transformers, as gMLP can achieve the same accuracy. For BERT, our model achieves parity with Transformers on pretraining perplexity and is better on some downstream tasks. On finetuning tasks where gMLP performs worse, making the gMLP model substantially larger can close the gap with Transformers. In general, our experiments show that gMLP can scale as well as Transformers over increased data and compute.

73.Unsupervised Deep Learning Methods for Biological Image Reconstruction ⬇️

Recently, deep learning approaches have become the main research frontier for biological image reconstruction problems thanks to their high performance, along with their ultra-fast reconstruction times. However, due to the difficulty of obtaining matched reference data for supervised learning, there has been increasing interest in unsupervised learning approaches that do not need paired reference data. In particular, self-supervised learning and generative models have been successfully used for various biological imaging applications. In this paper, we overview these approaches from a coherent perspective in the context of classical inverse problems, and discuss their applications to biological imaging.

74.DFENet: A Novel Dimension Fusion Edge Guided Network for Brain MRI Segmentation ⬇️

The rapid increment of morbidity of brain stroke in the last few years have been a driving force towards fast and accurate segmentation of stroke lesions from brain MRI images. With the recent development of deep-learning, computer-aided and segmentation methods of ischemic stroke lesions have been useful for clinicians in early diagnosis and treatment planning. However, most of these methods suffer from inaccurate and unreliable segmentation results because of their inability to capture sufficient contextual features from the MRI volumes. To meet these requirements, 3D convolutional neural networks have been proposed, which, however, suffer from huge computational requirements. To mitigate these problems, we propose a novel Dimension Fusion Edge-guided network (DFENet) that can meet both of these requirements by fusing the features of 2D and 3D CNNs. Unlike other methods, our proposed network uses a parallel partial decoder (PPD) module for aggregating and upsampling selected features, rich in important contextual information. Additionally, we use an edge-guidance and enhanced mixing loss for constantly supervising and improvising the learning process of the network. The proposed method is evaluated on publicly available Anatomical Tracings of Lesions After Stroke (ATLAS) dataset, resulting in mean DSC, IoU, Precision and Recall values of 0.5457, 0.4015, 0.6371, and 0.4969 respectively. The results, when compared to other state-of-the-art methods, outperforms them by a significant margin. Therefore, the proposed model is robust, accurate, superior to the existing methods, and can be relied upon for biomedical applications.

75.Joint Optimization of Hadamard Sensing and Reconstruction in Compressed Sensing Fluorescence Microscopy ⬇️

Compressed sensing fluorescence microscopy (CS-FM) proposes a scheme whereby less measurements are collected during sensing and reconstruction is performed to recover the image. Much work has gone into optimizing the sensing and reconstruction portions separately. We propose a method of jointly optimizing both sensing and reconstruction end-to-end under a total measurement constraint, enabling learning of the optimal sensing scheme concurrently with the parameters of a neural network-based reconstruction network. We train our model on a rich dataset of confocal, two-photon, and wide-field microscopy images comprising of a variety of biological samples. We show that our method outperforms several baseline sensing schemes and a regularized regression reconstruction algorithm.

76.CNN-based Approaches For Cross-Subject Classification in Motor Imagery: From The State-of-The-Art to DynamicNet ⬇️

Motor imagery (MI)-based brain-computer interface (BCI) systems are being increasingly employed to provide alternative means of communication and control for people suffering from neuro-motor impairments, with a special effort to bring these systems out of the controlled lab environments. Hence, accurately classifying MI from brain signals, e.g., from electroencephalography (EEG), is essential to obtain reliable BCI systems. However, MI classification is still a challenging task, because the signals are characterized by poor SNR, high intra-subject and cross-subject variability. Deep learning approaches have started to emerge as valid alternatives to standard machine learning techniques, e.g., filter bank common spatial pattern (FBCSP), to extract subject-independent features and to increase the cross-subject classification performance of MI BCI systems. In this paper, we first present a review of the most recent studies using deep learning for MI classification, with particular attention to their cross-subject performance. Second, we propose DynamicNet, a Python-based tool for quick and flexible implementations of deep learning models based on convolutional neural networks. We show-case the potentiality of DynamicNet by implementing EEGNet, a well-established architecture for effective EEG classification. Finally, we compare its performance with FBCSP in a 4-class MI classification over public datasets. To explore its cross-subject classification ability, we applied three different cross-validation schemes. From our results, we demonstrate that DynamicNet-implemented EEGNet outperforms FBCSP by about 25%, with a statistically significant difference when cross-subject validation schemes are applied.

77.A Review on Explainability in Multimodal Deep Neural Nets ⬇️

Artificial Intelligence techniques powered by deep neural nets have achieved much success in several application domains, most significantly and notably in the Computer Vision applications and Natural Language Processing tasks. Surpassing human-level performance propelled the research in the applications where different modalities amongst language, vision, sensory, text play an important role in accurate predictions and identification. Several multimodal fusion methods employing deep learning models are proposed in the literature. Despite their outstanding performance, the complex, opaque and black-box nature of the deep neural nets limits their social acceptance and usability. This has given rise to the quest for model interpretability and explainability, more so in the complex tasks involving multimodal AI methods. This paper extensively reviews the present literature to present a comprehensive survey and commentary on the explainability in multimodal deep neural nets, especially for the vision and language tasks. Several topics on multimodal AI and its applications for generic domains have been covered in this paper, including the significance, datasets, fundamental building blocks of the methods and techniques, challenges, applications, and future trends in this domain

78.Fast and Accurate Camera Scene Detection on Smartphones ⬇️

AI-powered automatic camera scene detection mode is nowadays available in nearly any modern smartphone, though the problem of accurate scene prediction has not yet been addressed by the research community. This paper for the first time carefully defines this problem and proposes a novel Camera Scene Detection Dataset (CamSDD) containing more than 11K manually crawled images belonging to 30 different scene categories. We propose an efficient and NPU-friendly CNN model for this task that demonstrates a top-3 accuracy of 99.5% on this dataset and achieves more than 200 FPS on the recent mobile SoCs. An additional in-the-wild evaluation of the obtained solution is performed to analyze its performance and limitation in the real-world scenarios. The dataset and pre-trained models used in this paper are available on the project website.

79.Real-Time Quantized Image Super-Resolution on Mobile NPUs, Mobile AI 2021 Challenge: Report ⬇️

Image super-resolution is one of the most popular computer vision problems with many important applications to mobile devices. While many solutions have been proposed for this task, they are usually not optimized even for common smartphone AI hardware, not to mention more constrained smart TV platforms that are often supporting INT8 inference only. To address this problem, we introduce the first Mobile AI challenge, where the target is to develop an end-to-end deep learning-based image super-resolution solutions that can demonstrate a real-time performance on mobile or edge NPUs. For this, the participants were provided with the DIV2K dataset and trained quantized models to do an efficient 3X image upscaling. The runtime of all models was evaluated on the Synaptics VS680 Smart Home board with a dedicated NPU capable of accelerating quantized neural networks. The proposed solutions are fully compatible with all major mobile AI accelerators and are capable of reconstructing Full HD images under 40-60 ms while achieving high fidelity results. A detailed description of all models developed in the challenge is provided in this paper.

80.Learned Smartphone ISP on Mobile NPUs with Deep Learning, Mobile AI 2021 Challenge: Report ⬇️

As the quality of mobile cameras starts to play a crucial role in modern smartphones, more and more attention is now being paid to ISP algorithms used to improve various perceptual aspects of mobile photos. In this Mobile AI challenge, the target was to develop an end-to-end deep learning-based image signal processing (ISP) pipeline that can replace classical hand-crafted ISPs and achieve nearly real-time performance on smartphone NPUs. For this, the participants were provided with a novel learned ISP dataset consisting of RAW-RGB image pairs captured with the Sony IMX586 Quad Bayer mobile sensor and a professional 102-megapixel medium format camera. The runtime of all models was evaluated on the MediaTek Dimensity 1000+ platform with a dedicated AI processing unit capable of accelerating both floating-point and quantized neural networks. The proposed solutions are fully compatible with the above NPU and are capable of processing Full HD photos under 60-100 milliseconds while achieving high fidelity results. A detailed description of all models developed in this challenge is provided in this paper.

81.Deep regression for uncertainty-aware and interpretable analysis of large-scale body MRI ⬇️

Large-scale medical studies such as the UK Biobank examine thousands of volunteer participants with medical imaging techniques. Combined with the vast amount of collected metadata, anatomical information from these images has the potential for medical analyses at unprecedented scale. However, their evaluation often requires manual input and long processing times, limiting the amount of reference values for biomarkers and other measurements available for research. Recent approaches with convolutional neural networks for regression can perform these evaluations automatically. On magnetic resonance imaging (MRI) data of more than 40,000 UK Biobank subjects, these systems can estimate human age, body composition and more. This style of analysis is almost entirely data-driven and no manual intervention or guidance with manually segmented ground truth images is required. The networks often closely emulate the reference method that provided their training data and can reach levels of agreement comparable to the expected variability between established medical gold standard techniques. The risk of silent failure can be individually quantified by predictive uncertainty obtained from a mean-variance criterion and ensembling. Saliency analysis furthermore enables an interpretation of the underlying relevant image features and showed that the networks learned to correctly target specific organs, limbs, and regions of interest.

82.A Cloud-based Deep Learning Framework for Remote Detection of Diabetic Foot Ulcers ⬇️

This research proposes a mobile and cloud-based framework for the automatic detection of diabetic foot ulcers and conducts an investigation of its performance. The system uses a cross-platform mobile framework which enables the deployment of mobile apps to multiple platforms using a single TypeScript code base. A deep convolutional neural network was deployed to a cloud-based platform where the mobile app could send photographs of patient's feet for inference to detect the presence of diabetic foot ulcers. The functionality and usability of the system were tested in two clinical settings: Salford Royal NHS Foundation Trust and Lancashire Teaching Hospitals NHS Foundation Trust. The benefits of the system, such as the potential use of the app by patients to identify and monitor their condition are discussed.

83.DOC3-Deep One Class Classification using Contradictions ⬇️

This paper introduces the notion of learning from contradictions (a.k.a Universum learning) for deep one class classification problems. We formalize this notion for the widely adopted one class large-margin loss, and propose the Deep One Class Classification using Contradictions (DOC3) algorithm. We show that learning from contradictions incurs lower generalization error by comparing the Empirical Radamacher Complexity (ERC) of DOC3 against its traditional inductive learning counterpart. Our empirical results demonstrate the efficacy of DOC3 algorithm achieving > 30% for CIFAR-10 and >50% for MV-Tec AD data sets in test AUCs compared to its inductive learning counterpart and in many cases improving the state-of-the-art in anomaly detection.

84.Disentangled Variational Information Bottleneck for Multiview Representation Learning ⬇️

Multiview data contain information from multiple modalities and have potentials to provide more comprehensive features for diverse machine learning tasks. A fundamental question in multiview analysis is what is the additional information brought by additional views and can quantitatively identify this additional information. In this work, we try to tackle this challenge by decomposing the entangled multiview features into shared latent representations that are common across all views and private representations that are specific to each single view. We formulate this feature disentanglement in the framework of information bottleneck and propose disentangled variational information bottleneck (DVIB). DVIB explicitly defines the properties of shared and private representations using constrains from mutual information. By deriving variational upper and lower bounds of mutual information terms, representations are efficiently optimized. We demonstrate the shared and private representations learned by DVIB well preserve the common labels shared between two views and unique labels corresponding to each single view, respectively. DVIB also shows comparable performance in classification task on images with corruptions. DVIB implementation is available at this https URL.

85.Dermoscopic Image Classification with Neural Style Transfer ⬇️

Skin cancer, the most commonly found human malignancy, is primarily diagnosed visually via dermoscopic analysis, biopsy, and histopathological examination. However, unlike other types of cancer, automated image classification of skin lesions is deemed more challenging due to the irregularity and variability in the lesions' appearances. In this work, we propose an adaptation of the Neural Style Transfer (NST) as a novel image pre-processing step for skin lesion classification problems. We represent each dermoscopic image as the style image and transfer the style of the lesion onto a homogeneous content image. This transfers the main variability of each lesion onto the same localized region, which allows us to integrate the generated images together and extract latent, low-rank style features via tensor decomposition. We train and cross-validate our model on a dermoscopic data set collected and preprocessed from the International Skin Imaging Collaboration (ISIC) database. We show that the classification performance based on the extracted tensor features using the style-transferred images significantly outperforms that of the raw images by more than 10%, and is also competitive with well-studied, pre-trained CNN models through transfer learning. Additionally, the tensor decomposition further identifies latent style clusters, which may provide clinical interpretation and insights.

86.Deep learning for detecting pulmonary tuberculosis via chest radiography: an international study across 10 countries ⬇️

Tuberculosis (TB) is a top-10 cause of death worldwide. Though the WHO recommends chest radiographs (CXRs) for TB screening, the limited availability of CXR interpretation is a barrier. We trained a deep learning system (DLS) to detect active pulmonary TB using CXRs from 9 countries across Africa, Asia, and Europe, and utilized large-scale CXR pretraining, attention pooling, and noisy student semi-supervised learning. Evaluation was on (1) a combined test set spanning China, India, US, and Zambia, and (2) an independent mining population in South Africa. Given WHO targets of 90% sensitivity and 70% specificity, the DLS's operating point was prespecified to favor sensitivity over specificity. On the combined test set, the DLS's ROC curve was above all 9 India-based radiologists, with an AUC of 0.90 (95%CI 0.87-0.92). The DLS's sensitivity (88%) was higher than the India-based radiologists (75% mean sensitivity), p<0.001 for superiority; and its specificity (79%) was non-inferior to the radiologists (84% mean specificity), p=0.004. Similar trends were observed within HIV positive and sputum smear positive sub-groups, and in the South Africa test set. We found that 5 US-based radiologists (where TB isn't endemic) were more sensitive and less specific than the India-based radiologists (where TB is endemic). The DLS also remained non-inferior to the US-based radiologists. In simulations, using the DLS as a prioritization tool for confirmatory testing reduced the cost per positive case detected by 40-80% compared to using confirmatory testing alone. To conclude, our DLS generalized to 5 countries, and merits prospective evaluation to assist cost-effective screening efforts in radiologist-limited settings. Operating point flexibility may permit customization of the DLS to account for site-specific factors such as TB prevalence, demographics, clinical resources, and customary practice patterns.

87.Private Facial Diagnosis as an Edge Service for Parkinson's DBS Treatment Valuation ⬇️

Facial phenotyping has recently been successfully exploited for medical diagnosis as a novel way to diagnose a range of diseases, where facial biometrics has been revealed to have rich links to underlying genetic or medical causes. In this paper, taking Parkinson's Diseases (PD) as a case study, we proposed an Artificial-Intelligence-of-Things (AIoT) edge-oriented privacy-preserving facial diagnosis framework to analyze the treatment of Deep Brain Stimulation (DBS) on PD patients. In the proposed framework, a new edge-based information theoretically secure framework is proposed to implement private deep facial diagnosis as a service over a privacy-preserving AIoT-oriented multi-party communication scheme, where partial homomorphic encryption (PHE) is leveraged to enable privacy-preserving deep facial diagnosis directly on encrypted facial patterns. In our experiments with a collected facial dataset from PD patients, for the first time, we demonstrated that facial patterns could be used to valuate the improvement of PD patients undergoing DBS treatment. We further implemented a privacy-preserving deep facial diagnosis framework that can achieve the same accuracy as the non-encrypted one, showing the potential of our privacy-preserving facial diagnosis as an trustworthy edge service for grading the severity of PD in patients.

88.Advances in Artificial Intelligence to Reduce Polyp Miss Rates during Colonoscopy ⬇️

BACKGROUND AND CONTEXT: Artificial intelligence has the potential to aid gastroenterologists by reducing polyp miss detection rates during colonoscopy screening for colorectal cancer.
NEW FINDINGS: We introduce a new deep neural network architecture, the Focus U-Net, which achieves state-of-the-art performance for polyp segmentation across five public datasets containing images of polyps obtained during colonoscopy.
LIMITATIONS: The model has been validated on images taken during colonoscopy but requires validation on live video data to ensure generalisability.
IMPACT: Once validated on live video data, our polyp segmentation algorithm could be integrated into colonoscopy practice and assist gastroenterologists by reducing the number of polyps missed

89.MSRF-Net: A Multi-Scale Residual Fusion Network for Biomedical Image Segmentation ⬇️

Methods based on convolutional neural networks have improved the performance of biomedical image segmentation. However, most of these methods cannot efficiently segment objects of variable sizes and train on small and biased datasets, which are common in biomedical use cases. While methods exist that incorporate multi-scale fusion approaches to address the challenges arising with variable sizes, they usually use complex models that are more suitable for general semantic segmentation computer vision problems. In this paper, we propose a novel architecture called MSRF-Net, which is specially designed for medical image segmentation tasks. The proposed MSRF-Net is able to exchange multi-scale features of varying receptive fields using a dual-scale dense fusion block (DSDF). Our DSDF block can exchange information rigorously across two different resolution scales, and our MSRF sub-network uses multiple DSDF blocks in sequence to perform multi-scale fusion. This allows the preservation of resolution, improved information flow, and propagation of both high- and low-level features to obtain accurate segmentation maps. The proposed MSRF-Net allows to capture object variabilities and provides improved results on different biomedical datasets. Extensive experiments on MSRF-Net demonstrate that the proposed method outperforms most of the cutting-edge medical image segmentation state-of-the-art methods. MSRF-Net advances the performance on four publicly available datasets, and also, MSRF-Net is more generalizable as compared to state-of-the-art methods.

90.Unsupervised MMRegNet based on Spatially Encoded Gradient Information ⬇️

Multi-modality medical images can provide relevant and complementary anatomical information for a target (organ, tumor or tissue). Registering the multi-modality images to a common space can fuse these comprehensive information, and bring convenience for clinical application. Recently, neural networks have been widely investigated to boost registration methods. However, it is still challenging to develop a multi-modality registration network due to the lack of robust criteria for network training. Besides, most existing registration networks mainly focus on pairwise registration, and can hardly be applicable for multiple image scenarios. In this work, we propose a multi-modality registration network (MMRegNet), which can jointly register multiple images with different modalities to a target image. Meanwhile, we present spatially encoded gradient information to train the MMRegNet in an unsupervised manner. The proposed network was evaluated on two datasets, i.e, MM-WHS 2017 and CHAOS 2019. The results show that the proposed network can achieve promising performance for cardiac left ventricle and liver registration tasks. Source code is released publicly on github.

91.Expressive Explanations of DNNs by Combining Concept Analysis with ILP ⬇️

Explainable AI has emerged to be a key component for black-box machine learning approaches in domains with a high demand for reliability or transparency. Examples are medical assistant systems, and applications concerned with the General Data Protection Regulation of the European Union, which features transparency as a cornerstone. Such demands require the ability to audit the rationale behind a classifier's decision. While visualizations are the de facto standard of explanations, they come short in terms of expressiveness in many ways: They cannot distinguish between different attribute manifestations of visual features (e.g. eye open vs. closed), and they cannot accurately describe the influence of absence of, and relations between features. An alternative would be more expressive symbolic surrogate models. However, these require symbolic inputs, which are not readily available in most computer vision tasks. In this paper we investigate how to overcome this: We use inherent features learned by the network to build a global, expressive, verbal explanation of the rationale of a feed-forward convolutional deep neural network (DNN). The semantics of the features are mined by a concept analysis approach trained on a set of human understandable visual concepts. The explanation is found by an Inductive Logic Programming (ILP) method and presented as first-order rules. We show that our explanation is faithful to the original black-box model.
The code for our experiments is available at this https URL.

92.Real-time Detection of Practical Universal Adversarial Perturbations ⬇️

Universal Adversarial Perturbations (UAPs) are a prominent class of adversarial examples that exploit the systemic vulnerabilities and enable physically realizable and robust attacks against Deep Neural Networks (DNNs). UAPs generalize across many different inputs; this leads to realistic and effective attacks that can be applied at scale. In this paper we propose HyperNeuron, an efficient and scalable algorithm that allows for the real-time detection of UAPs by identifying suspicious neuron hyper-activations. Our results show the effectiveness of HyperNeuron on multiple tasks (image classification, object detection), against a wide variety of universal attacks, and in realistic scenarios, like perceptual ad-blocking and adversarial patches. HyperNeuron is able to simultaneously detect both adversarial mask and patch UAPs with comparable or better performance than existing UAP defenses whilst introducing a significantly reduced latency of only 0.86 milliseconds per image. This suggests that many realistic and practical universal attacks can be reliably mitigated in real-time, which shows promise for the robust deployment of machine learning systems.

93.Is In-Domain Data Really Needed? A Pilot Study on Cross-Domain Calibration for Network Quantization ⬇️

Post-training quantization methods use a set of calibration data to compute quantization ranges for network parameters and activations. The calibration data usually comes from the training dataset which could be inaccessible due to sensitivity of the data. In this work, we want to study such a problem: can we use out-of-domain data to calibrate the trained networks without knowledge of the original dataset? Specifically, we go beyond the domain of natural images to include drastically different domains such as X-ray images, satellite images and ultrasound images. We find cross-domain calibration leads to surprisingly stable performance of quantized models on 10 tasks in different image domains with 13 different calibration datasets. We also find that the performance of quantized models is correlated with the similarity of the Gram matrices between the source and calibration domains, which can be used as a criterion to choose calibration set for better performance. We believe our research opens the door to borrow cross-domain knowledge for network quantization and compression.

94.Texture Generation with Neural Cellular Automata ⬇️

Neural Cellular Automata (NCA) have shown a remarkable ability to learn the required rules to "grow" images, classify morphologies, segment images, as well as to do general computation such as path-finding. We believe the inductive prior they introduce lends itself to the generation of textures. Textures in the natural world are often generated by variants of locally interacting reaction-diffusion systems. Human-made textures are likewise often generated in a local manner (textile weaving, for instance) or using rules with local dependencies (regular grids or geometric patterns). We demonstrate learning a texture generator from a single template image, with the generation method being embarrassingly parallel, exhibiting quick convergence and high fidelity of output, and requiring only some minimal assumptions around the underlying state manifold. Furthermore, we investigate properties of the learned models that are both useful and interesting, such as non-stationary dynamics and an inherent robustness to damage. Finally, we make qualitative claims that the behaviour exhibited by the NCA model is a learned, distributed, local algorithm to generate a texture, setting our method apart from existing work on texture generation. We discuss the advantages of such a paradigm.

95.Neural Trees for Learning on Graphs ⬇️

Graph Neural Networks (GNNs) have emerged as a flexible and powerful approach for learning over graphs. Despite this success, existing GNNs are constrained by their local message-passing architecture and are provably limited in their expressive power. In this work, we propose a new GNN architecture -- the Neural Tree. The neural tree architecture does not perform message passing on the input graph but on a tree-structured graph, called the H-tree, that is constructed from the input graph. Nodes in the H-tree correspond to subgraphs in the input graph, and they are reorganized in a hierarchical manner such that a parent-node of a node in the H-tree always corresponds to a larger subgraph in the input graph. We show that the neural tree architecture can approximate any smooth probability distribution function over an undirected graph, as well as emulate the junction tree algorithm. We also prove that the number of parameters needed to achieve an $\epsilon$-approximation of the distribution function is exponential in the treewidth of the input graph, but linear in its size. We apply the neural tree to semi-supervised node classification in 3D scene graphs, and show that these theoretical properties translate into significant gains in prediction accuracy, over the more traditional GNN architectures.

96.Rethinking Skip Connection with Layer Normalization in Transformers and ResNets ⬇️

Skip connection, is a widely-used technique to improve the performance and the convergence of deep neural networks, which is believed to relieve the difficulty in optimization due to non-linearity by propagating a linear component through the neural network layers. However, from another point of view, it can also be seen as a modulating mechanism between the input and the output, with the input scaled by a pre-defined value one. In this work, we investigate how the scale factors in the effectiveness of the skip connection and reveal that a trivial adjustment of the scale will lead to spurious gradient exploding or vanishing in line with the deepness of the models, which could be addressed by normalization, in particular, layer normalization, which induces consistent improvements over the plain skip connection. Inspired by the findings, we further propose to adaptively adjust the scale of the input by recursively applying skip connection with layer normalization, which promotes the performance substantially and generalizes well across diverse tasks including both machine translation and image classification datasets.

97.Multi-scale super-resolution generation of low-resolution scanned pathological images ⬇️

Digital pathology slide is easy to store and manage, convenient to browse and transmit. However, because of the high-resolution scan for example 40 times magnification(40X) during the digitization, the file size of each whole slide image exceeds 1Gigabyte, which eventually leads to huge storage capacity and very slow network transmission. We design a strategy to scan slides with low resolution (5X) and a super-resolution method is proposed to restore the image details when in diagnosis. The method is based on a multi-scale generative adversarial network, which sequentially generate three high-resolution images such as 10X, 20X and 40X. The perceived loss, generator loss of the generated images and real images are compared on three image resolutions, and a discriminator is used to evaluate the difference of highest-resolution generated image and real image. A dataset consisting of 100,000 pathological images from 10 types of human tissues is performed for training and testing the network. The generated images have high peak-signal-to-noise-ratio (PSNR) and structural-similarity-index (SSIM). The PSNR of 10X to 40X image are 24.16, 22.27 and 20.44, and the SSIM are 0.845, 0.680 and 0.512, which are better than other super-resolution networks such as DBPN, ESPCN, RDN, EDSR and MDSR. Moreover, visual inspections show that the generated high-resolution images by our network have enough details for diagnosis, good color reproduction and close to real images, while other five networks are severely blurred, local deformation or miss important details. Moreover, no significant differences can be found on pathological diagnosis based on the generated and real images. The proposed multi-scale network can generate good high-resolution pathological images, and will provide a low-cost storage (about 15MB/image on 5X), faster image sharing method for digital pathology.

98.Make Bipedal Robots Learn How to Imitate ⬇️

Bipedal robots do not perform well as humans since they do not learn to walk like we do. In this paper we propose a method to train a bipedal robot to perform some basic movements with the help of imitation learning (IL) in which an instructor will perform the movement and the robot will try to mimic the instructor movement. To the best of our knowledge, this is the first time we train the robot to perform movements with a single video of the instructor and as the training is done based on joint angles the robot will keep its joint angles always in physical limits which in return help in faster training. The joints of the robot are identified by OpenPose architecture and then joint angle data is extracted with the help of angle between three points resulting in a noisy solution. We smooth the data using Savitzky-Golay filter and preserve the Simulatore data anatomy. An ingeniously written Deep Q Network (DQN) is trained with experience replay to make the robot learn to perform the movements as similar as the instructor. The implementation of the paper is made publicly available.

99.Window-Level is a Strong Denoising Surrogate ⬇️

CT image quality is heavily reliant on radiation dose, which causes a trade-off between radiation dose and image quality that affects the subsequent image-based diagnostic performance. However, high radiation can be harmful to both patients and operators. Several (deep learning-based) approaches have been attempted to denoise low dose images. However, those approaches require access to large training sets, specifically the full dose CT images for reference, which can often be difficult to obtain. Self-supervised learning is an emerging alternative for lowering the reference data requirement facilitating unsupervised learning. Currently available self-supervised CT denoising works are either dependent on foreign domain or pretexts are not very task-relevant. To tackle the aforementioned challenges, we propose a novel self-supervised learning approach, namely Self-Supervised Window-Leveling for Image DeNoising (SSWL-IDN), leveraging an innovative, task-relevant, simple, yet effective surrogate -- prediction of the window-leveled equivalent. SSWL-IDN leverages residual learning and a hybrid loss combining perceptual loss and MSE, all incorporated in a VAE framework. Our extensive (in- and cross-domain) experimentation demonstrates the effectiveness of SSWL-IDN in aggressive denoising of CT (abdomen and chest) images acquired at 5% dose level only.

100.RIDnet: Radiologist-Inspired Deep Neural Network for Low-dose CT Denoising ⬇️

Being low-level radiation exposure and less harmful to health, low-dose computed tomography (LDCT) has been widely adopted in the early screening of lung cancer and COVID-19. LDCT images inevitably suffer from the degradation problem caused by complex noises. It was reported that, compared with commercial iterative reconstruction methods, deep learning (DL)-based LDCT denoising methods using convolutional neural network (CNN) achieved competitive performance. Most existing DL-based methods focus on the local information extracted by CNN, while ignoring both explicit non-local and context information (which are leveraged by radiologists). To address this issue, we propose a novel deep learning model named radiologist-inspired deep denoising network (RIDnet) to imitate the workflow of a radiologist reading LDCT images. Concretely, the proposed model explicitly integrates all the local, non-local and context information rather than local information only. Our radiologist-inspired model is potentially favoured by radiologists as a familiar workflow. A double-blind reader study on a public clinical dataset shows that, compared with state-of-the-art methods, our proposed model achieves the most impressive performance in terms of the structural fidelity, the noise suppression and the overall score. As a physicians-inspired model, RIDnet gives a new research roadmap that takes into account the behavior of physicians when designing decision support tools for assisting clinical diagnosis. Models and code are available at this https URL.

101.NeuroGen: activation optimized image synthesis for discovery neuroscience ⬇️

Functional MRI (fMRI) is a powerful technique that has allowed us to characterize visual cortex responses to stimuli, yet such experiments are by nature constructed based on a priori hypotheses, limited to the set of images presented to the individual while they are in the scanner, are subject to noise in the observed brain responses, and may vary widely across individuals. In this work, we propose a novel computational strategy, which we call NeuroGen, to overcome these limitations and develop a powerful tool for human vision neuroscience discovery. NeuroGen combines an fMRI-trained neural encoding model of human vision with a deep generative network to synthesize images predicted to achieve a target pattern of macro-scale brain activation. We demonstrate that the reduction of noise that the encoding model provides, coupled with the generative network's ability to produce images of high fidelity, results in a robust discovery architecture for visual neuroscience. By using only a small number of synthetic images created by NeuroGen, we demonstrate that we can detect and amplify differences in regional and individual human brain response patterns to visual stimuli. We then verify that these discoveries are reflected in the several thousand observed image responses measured with fMRI. We further demonstrate that NeuroGen can create synthetic images predicted to achieve regional response patterns not achievable by the best-matching natural images. The NeuroGen framework extends the utility of brain encoding models and opens up a new avenue for exploring, and possibly precisely controlling, the human visual system.

102.Image Super-Resolution Quality Assessment: Structural Fidelity Versus Statistical Naturalness ⬇️

Single image super-resolution (SISR) algorithms reconstruct high-resolution (HR) images with their low-resolution (LR) counterparts. It is desirable to develop image quality assessment (IQA) methods that can not only evaluate and compare SISR algorithms, but also guide their future development. In this paper, we assess the quality of SISR generated images in a two-dimensional (2D) space of structural fidelity versus statistical naturalness. This allows us to observe the behaviors of different SISR algorithms as a tradeoff in the 2D space. Specifically, SISR methods are traditionally designed to achieve high structural fidelity but often sacrifice statistical naturalness, while recent generative adversarial network (GAN) based algorithms tend to create more natural-looking results but lose significantly on structural fidelity. Furthermore, such a 2D evaluation can be easily fused to a scalar quality prediction. Interestingly, we find that a simple linear combination of a straightforward local structural fidelity and a global statistical naturalness measures produce surprisingly accurate predictions of SISR image quality when tested using public subject-rated SISR image datasets. Code of the proposed SFSN model is publicly available at \url{this https URL}.

103.High-Robustness, Low-Transferability Fingerprinting of Neural Networks ⬇️

This paper proposes Characteristic Examples for effectively fingerprinting deep neural networks, featuring high-robustness to the base model against model pruning as well as low-transferability to unassociated models. This is the first work taking both robustness and transferability into consideration for generating realistic fingerprints, whereas current methods lack practical assumptions and may incur large false positive rates. To achieve better trade-off between robustness and transferability, we propose three kinds of characteristic examples: vanilla C-examples, RC-examples, and LTRC-example, to derive fingerprints from the original base model. To fairly characterize the trade-off between robustness and transferability, we propose Uniqueness Score, a comprehensive metric that measures the difference between robustness and transferability, which also serves as an indicator to the false alarm problem.

104.SA-GAN: Structure-Aware Generative Adversarial Network for Shape-Preserving Synthetic CT Generation ⬇️

In medical image synthesis, model training could be challenging due to the inconsistencies between images of different modalities even with the same patient, typically caused by internal status/tissue changes as different modalities are usually obtained at a different time. This paper proposes a novel deep learning method, Structure-aware Generative Adversarial Network (SA-GAN), that preserves the shapes and locations of in-consistent structures when generating medical images. SA-GAN is employed to generate synthetic computed tomography (synCT) images from magnetic resonance imaging (MRI) with two parallel streams: the global stream translates the input from the MRI to the CT domain while the local stream automatically segments the inconsistent organs, maintains their locations and shapes in MRI, and translates the organ intensities to CT. Through extensive experiments on a pelvic dataset, we demonstrate that SA-GAN provides clinically acceptable accuracy on both synCTs and organ segmentation and supports MR-only treatment planning in disease sites with internal organ status changes.

105.Learning a Universal Template for Few-shot Dataset Generalization ⬇️

Few-shot dataset generalization is a challenging variant of the well-studied few-shot classification problem where a diverse training set of several datasets is given, for the purpose of training an adaptable model that can then learn classes from new datasets using only a few examples. To this end, we propose to utilize the diverse training set to construct a universal template: a partial model that can define a wide array of dataset-specialized models, by plugging in appropriate components. For each new few-shot classification problem, our approach therefore only requires inferring a small number of parameters to insert into the universal template. We design a separate network that produces an initialization of those parameters for each given task, and we then fine-tune its proposed initialization via a few steps of gradient descent. Our approach is more parameter-efficient, scalable and adaptable compared to previous methods, and achieves the state-of-the-art on the challenging Meta-Dataset benchmark.

106.Volumetric Instance-Aware Semantic Mapping and 3D Object Discovery ⬇️

To autonomously navigate and plan interactions in real-world environments, robots require the ability to robustly perceive and map complex, unstructured surrounding scenes. Besides building an internal representation of the observed scene geometry, the key insight toward a truly functional understanding of the environment is the usage of higher-level entities during mapping, such as individual object instances. We propose an approach to incrementally build volumetric object-centric maps during online scanning with a localized RGB-D camera. First, a per-frame segmentation scheme combines an unsupervised geometric approach with instance-aware semantic object predictions. This allows us to detect and segment elements both from the set of known classes and from other, previously unseen categories. Next, a data association step tracks the predicted instances across the different frames. Finally, a map integration strategy fuses information about their 3D shape, location, and, if available, semantic class into a global volume. Evaluation on a publicly available dataset shows that the proposed approach for building instance-level semantic maps is competitive with state-of-the-art methods, while additionally able to discover objects of unseen categories. The system is further evaluated within a real-world robotic mapping setup, for which qualitative results highlight the online nature of the method.