20201013.md

ArXiv cs.CV --Tue, 13 Oct 2020

1.Cut-and-Paste Neural Rendering ⬇️

Cut-and-paste methods take an object from one image and insert it into another. Doing so often results in unrealistic looking images because the inserted object's shading is inconsistent with the target scene's shading. Existing reshading methods require a geometric and physical model of the inserted object, which is then rendered using environment parameters. Accurately constructing such a model only from a single image is beyond the current understanding of computer vision. We describe an alternative procedure -- cut-and-paste neural rendering, to render the inserted fragment's shading field consistent with the target scene. We use a Deep Image Prior (DIP) as a neural renderer trained to render an image with consistent image decomposition inferences. The resulting rendering from DIP should have an albedo consistent with composite albedo; it should have a shading field that, outside the inserted fragment, is the same as the target scene's shading field; and composite surface normals are consistent with the final rendering's shading field. The result is a simple procedure that produces convincing and realistic shading. Moreover, our procedure does not require rendered images or image-decomposition from real images in the training or labeled annotations. In fact, our only use of simulated ground truth is our use of a pre-trained normal estimator. Qualitative results are strong, supported by a user study comparing against the state-of-the-art image harmonization baseline.

2.PANDA -- Adapting Pretrained Features for Anomaly Detection ⬇️

Anomaly detection methods require high-quality features. One way of obtaining strong features is to adapt pre-trained features to anomaly detection on the target distribution. Unfortunately, simple adaptation methods often result in catastrophic collapse (feature deterioration) and reduce performance. DeepSVDD combats collapse by removing biases from architectures, but this limits the adaptation performance gain. In this work, we propose two methods for combating collapse: i) a variant of early stopping that dynamically learns the stopping iteration ii) elastic regularization inspired by continual learning. In addition, we conduct a thorough investigation of Imagenet-pretrained features for one-class anomaly detection. Our method, PANDA, outperforms the state-of-the-art in the one-class and outlier exposure settings (CIFAR10: 96.2% vs. 90.1% and 98.9% vs. 95.6%).

3.Webly Supervised Image Classification with Metadata: Automatic Noisy Label Correction via Visual-Semantic Graph ⬇️

Webly supervised learning becomes attractive recently for its efficiency in data expansion without expensive human labeling. However, adopting search queries or hashtags as web labels of images for training brings massive noise that degrades the performance of DNNs. Especially, due to the semantic confusion of query words, the images retrieved by one query may contain tremendous images belonging to other concepts. For example, searching `tiger cat' on Flickr will return a dominating number of tiger images rather than the cat images. These realistic noisy samples usually have clear visual semantic clusters in the visual space that mislead DNNs from learning accurate semantic labels. To correct real-world noisy labels, expensive human annotations seem indispensable. Fortunately, we find that metadata can provide extra knowledge to discover clean web labels in a labor-free fashion, making it feasible to automatically provide correct semantic guidance among the massive label-noisy web data. In this paper, we propose an automatic label corrector VSGraph-LC based on the visual-semantic graph. VSGraph-LC starts from anchor selection referring to the semantic similarity between metadata and correct label concepts, and then propagates correct labels from anchors on a visual graph using graph neural network (GNN). Experiments on realistic webly supervised learning datasets Webvision-1000 and NUS-81-Web show the effectiveness and robustness of VSGraph-LC. Moreover, VSGraph-LC reveals its advantage on the open-set validation set.

4.On the Minimal Recognizable Image Patch ⬇️

In contrast to human vision, common recognition algorithms often fail on partially occluded images. We propose characterizing, empirically, the algorithmic limits by finding a minimal recognizable patch (MRP) that is by itself sufficient to recognize the image. A specialized deep network allows us to find the most informative patches of a given size, and serves as an experimental tool. A human vision study recently characterized related (but different) minimally recognizable configurations (MIRCs) [1], for which we specify computational analogues (denoted cMIRCs). The drop in human decision accuracy associated with size reduction of these MIRCs is substantial and sharp. Interestingly, such sharp reductions were also found for the computational versions we specified.

5.Fully Automatic Wound Segmentation with Deep Convolutional Neural Networks ⬇️

Acute and chronic wounds have varying etiologies and are an economic burden to healthcare systems around the world. The advanced wound care market is expected to exceed $22 billion by 2024. Wound care professionals rely heavily on images and image documentation for proper diagnosis and treatment. Unfortunately lack of expertise can lead to improper diagnosis of wound etiology and inaccurate wound management and documentation. Fully automatic segmentation of wound areas in natural images is an important part of the diagnosis and care protocol since it is crucial to measure the area of the wound and provide quantitative parameters in the treatment. Various deep learning models have gained success in image analysis including semantic segmentation. Particularly, MobileNetV2 stands out among others due to its lightweight architecture and uncompromised performance. This manuscript proposes a novel convolutional framework based on MobileNetV2 and connected component labelling to segment wound regions from natural images. We build an annotated wound image dataset consisting of 1,109 foot ulcer images from 889 patients to train and test the deep learning models. We demonstrate the effectiveness and mobility of our method by conducting comprehensive experiments and analyses on various segmentation neural networks.

6.Neural Enhancement in Content Delivery Systems: The State-of-the-Art and Future Directions ⬇️

Internet-enabled smartphones and ultra-wide displays are transforming a variety of visual apps spanning from on-demand movies and 360-degree videos to video-conferencing and live streaming. However, robustly delivering visual content under fluctuating networking conditions on devices of diverse capabilities remains an open problem. In recent years, advances in the field of deep learning on tasks such as super-resolution and image enhancement have led to unprecedented performance in generating high-quality images from low-quality ones, a process we refer to as neural enhancement. In this paper, we survey state-of-the-art content delivery systems that employ neural enhancement as a key component in achieving both fast response time and high visual quality. We first present the deployment challenges of neural enhancement models. We then cover systems targeting diverse use-cases and analyze their design decisions in overcoming technical challenges. Moreover, we present promising directions based on the latest insights from deep learning research to further boost the quality of experience of these systems.

7.Viewpoint-Aware Channel-Wise Attentive Network for Vehicle Re-Identification ⬇️

Vehicle re-identification (re-ID) matches images of the same vehicle across different cameras. It is fundamentally challenging because the dramatically different appearance caused by different viewpoints would make the framework fail to match two vehicles of the same identity. Most existing works solved the problem by extracting viewpoint-aware feature via spatial attention mechanism, which, yet, usually suffers from noisy generated attention map or otherwise requires expensive keypoint labels to improve the quality. In this work, we propose Viewpoint-aware Channel-wise Attention Mechanism (VCAM) by observing the attention mechanism from a different aspect. Our VCAM enables the feature learning framework channel-wisely reweighing the importance of each feature maps according to the "viewpoint" of input vehicle. Extensive experiments validate the effectiveness of the proposed method and show that we perform favorably against state-of-the-arts methods on the public VeRi-776 dataset and obtain promising results on the 2020 AI City Challenge. We also conduct other experiments to demonstrate the interpretability of how our VCAM practically assists the learning framework.

8.Pedestrian Trajectory Prediction with Convolutional Neural Networks ⬇️

Predicting the future trajectories of pedestrians is a challenging problem that has a range of application, from crowd surveillance to autonomous driving. In literature, methods to approach pedestrian trajectory prediction have evolved, transitioning from physics-based models to data-driven models based on recurrent neural networks. In this work, we propose a new approach to pedestrian trajectory prediction, with the introduction of a novel 2D convolutional model. This new model outperforms recurrent models, and it achieves state-of-the-art results on the ETH and TrajNet datasets. We also present an effective system to represent pedestrian positions and powerful data augmentation techniques, such as the addition of Gaussian noise and the use of random rotations, which can be applied to any model. As an additional exploratory analysis, we present experimental results on the inclusion of occupancy methods to model social information, which empirically show that these methods are ineffective in capturing social interaction.

9.Permuted AdaIN: Enhancing the Representation of Local Cues in Image Classifiers ⬇️

Recent work has shown that convolutional neural network classifiers overly rely on texture at the expense of shape cues, which adversely affects the classifier's performance in shifted domains. In this work, we make a similar but different distinction between local image cues, including shape and texture, and global image statistics. We provide a method that enhances the representation of local cues in the hidden layers of image classifiers. Our method, called Permuted Adaptive Instance Normalization (pAdaIN), samples a random permutation $\pi$ that rearranges the samples in a given batch. Adaptive Instance Normalization (AdaIN) is then applied between the activations of each (non-permuted) sample $i$ and the corresponding activations of the sample $\pi(i)$, thus swapping statistics between the samples of the batch. Since the global image statistics are distorted, this swapping procedure causes the network to rely on the local image cues. By choosing the random permutation with probability $p$ and the identity permutation otherwise, one can control the strength of this effect. With the correct choice of $p$, selected without considering the test data, our method consistently outperforms baseline methods in image classification, as well as in the setting of domain generalization.

10.Hierarchical Attention Learning of Scene Flow in 3D Point Clouds ⬇️

Scene flow represents the 3D motion of every point in the dynamic environments. Like the optical flow that represents the motion of pixels in 2D images, 3D motion representation of scene flow benefits many applications, such as autonomous driving and service robot. This paper studies the problem of scene flow estimation from two consecutive 3D point clouds. In this paper, a novel hierarchical neural network with double attention is proposed for learning the correlation of point features in adjacent frames and refining scene flow from coarse to fine layer by layer. The proposed network has a new more-for-less hierarchical architecture. The more-for-less means that the number of input points is greater than the number of output points for scene flow estimation, which brings more input information and balances the precision and resource consumption. In this hierarchical architecture, scene flow of different levels is generated and supervised respectively. A novel attentive embedding module is introduced to aggregate the features of adjacent points using a double attention method in a patch-to-patch manner. The proper layers for flow embedding and flow supervision are carefully considered in our network designment. Experiments show that the proposed network outperforms the state-of-the-art performance of 3D scene flow estimation on the FlyingThings3D and KITTI Scene Flow 2015 datasets. We also apply the proposed network to realistic LiDAR odometry task, which is an key problem in autonomous driving. The experiment results demonstrate that our proposed network can outperform the ICP-based method and shows the good practical application ability.

11.Unsupervised Image-to-Image Translation via Pre-trained StyleGAN2 Network ⬇️

Image-to-Image (I2I) translation is a heated topic in academia, and it also has been applied in real-world industry for tasks like image synthesis, super-resolution, and colorization. However, traditional I2I translation methods train data in two or more domains together. This requires lots of computation resources. Moreover, the results are of lower quality, and they contain many more artifacts. The training process could be unstable when the data in different domains are not balanced, and modal collapse is more likely to happen. We proposed a new I2I translation method that generates a new model in the target domain via a series of model transformations on a pre-trained StyleGAN2 model in the source domain. After that, we proposed an inversion method to achieve the conversion between an image and its latent vector. By feeding the latent vector into the generated model, we can perform I2I translation between the source domain and target domain. Both qualitative and quantitative evaluations were conducted to prove that the proposed method can achieve outstanding performance in terms of image quality, diversity and semantic similarity to the input and reference images compared to state-of-the-art works.

12.A Novel Strategy for COVID-19 Classification from Chest X-ray Images Using Deep Stacked-Ensembles ⬇️

The issue of COVID-19, increasing notably with a massive mortality rate has led to the WHO declaring it as a pandemic. The unavailability of an antiviral drug has also led to dismay world-wide. The diagnostic tests are performed by collecting samples inserting a swab into the nasal or oral cavity. These collected samples and then sent to laboratories for viral-tests. Recently, chest radiographs are used to observe the proportion of virus in the lungs at various section-scans. As laboratory testing is time-consuming with a lot of human effort, diagnosis using chest radiographs is considered to be a befitting choice. In this research, a stacked-ensemble model is designed to classify whether a patient is infected with COVID-19, viral-pneumonia or has a healthy profile by considering chest X-ray images. A lot of complications were observed from existing literature in classifying COVID-19 radiographic images and are eliminated using our methodology. A training algorithm is constructed to speed up the training process which acquired good generalisations. The proposed model resulted in state-of-the-art outcomes with an accuracy score of 99.48% for binary classification and 97.4% for tri-class classification. Additionally, visualisations are illustrated for a detailed comprehension of the model's perception for the information provided to it.

13.Asymmetric Siamese Networks for Semantic Change Detection ⬇️

Given two multi-temporal aerial images, semantic change detection aims to locate the land-cover variations and identify their categories with pixel-wise boundaries. The problem has demonstrated promising potentials in many earth vision related tasks, such as precise urban planning and natural resource management. Existing state-of-the-art algorithms mainly identify the changed pixels through symmetric modules, which would suffer from categorical ambiguity caused by changes related to totally different land-cover distributions. In this paper, we present an asymmetric siamese network (ASN) to locate and identify semantic changes through feature pairs obtained from modules of widely different structures, which involve different spatial ranges and quantities of parameters to factor in the discrepancy across different land-cover distributions. To better train and evaluate our model, we create a large-scale well-annotated SEmantic Change detectiON Dataset (SECOND), while an adaptive threshold learning (ATL) module and a separated kappa (SeK) coefficient are proposed to alleviate the influences of label imbalance in model training and evaluation. The experimental results demonstrate that the proposed model can stably outperform the state-of-the-art algorithms with different encoder backbones.

14.Graph Regularized Nonnegative Tensor Ring Decomposition for Multiway Representation Learning ⬇️

Tensor ring (TR) decomposition is a powerful tool for exploiting the low-rank nature of multiway data and has demonstrated great potential in a variety of important applications. In this paper, nonnegative tensor ring (NTR) decomposition and graph regularized NTR (GNTR) decomposition are proposed, where the former equips TR decomposition with local feature extraction by imposing nonnegativity on the core tensors and the latter is additionally able to capture manifold geometry information of tensor data, both significantly extend the applications of TR decomposition for nonnegative multiway representation learning. Accelerated proximal gradient based methods are derived for NTR and GNTR. The experimental result demonstrate that the proposed algorithms can extract parts-based basis with rich colors and rich lines from tensor objects that provide more interpretable and meaningful representation, and hence yield better performance than the state-of-the-art tensor based methods in clustering and classification tasks.

15.The MECCANO Dataset: Understanding Human-Object Interactions from Egocentric Videos in an Industrial-like Domain ⬇️

Wearable cameras allow to collect images and videos of humans interacting with the world. While human-object interactions have been thoroughly investigated in third person vision, the problem has been understudied in egocentric settings and in industrial scenarios. To fill this gap, we introduce MECCANO, the first dataset of egocentric videos to study human-object interactions in industrial-like settings. MECCANO has been acquired by 20 participants who were asked to build a motorbike model, for which they had to interact with tiny objects and tools. The dataset has been explicitly labeled for the task of recognizing human-object interactions from an egocentric perspective. Specifically, each interaction has been labeled both temporally (with action segments) and spatially (with active object bounding boxes). With the proposed dataset, we investigate four different tasks including 1) action recognition, 2) active object detection, 3) active object recognition and 4) egocentric human-object interaction detection, which is a revisited version of the standard human-object interaction detection task. Baseline results show that the MECCANO dataset is a challenging benchmark to study egocentric human-object interactions in industrial-like scenarios. We publicy release the dataset at this https URL.

16.Convolutional Neural Network optimization via Channel Reassessment Attention module ⬇️

The performance of convolutional neural networks (CNNs) can be improved by adjusting the interrelationship between channels with attention mechanism. However, attention mechanism in recent advance has not fully utilized spatial information of feature maps, which makes a great difference to the results of generated channel attentions. In this paper, we propose a novel network optimization module called Channel Reassessment Attention (CRA) module which uses channel attentions with spatial information of feature maps to enhance representational power of networks. We employ CRA module to assess channel attentions based on feature maps in different channels, then the final features are refined adaptively by product between channel attentions and feature maps.CRA module is a computational lightweight module and it can be embedded into any architectures of CNNs. The experiments on ImageNet, CIFAR and MS COCO datasets demonstrate that the embedding of CRA module on various networks effectively improves the performance under different evaluation standards.

17.Omni-Directional Image Generation from Single Snapshot Image ⬇️

An omni-directional image (ODI) is the image that has a field of view covering the entire sphere around the camera. The ODIs have begun to be used in a wide range of fields such as virtual reality (VR), robotics, and social network services. Although the contents using ODI have increased, the available images and videos are still limited, compared with widespread snapshot images. A large number of ODIs are desired not only for the VR contents, but also for training deep learning models for ODI. For these purposes, a novel computer vision task to generate ODI from a single snapshot image is proposed in this paper. To tackle this problem, the conditional generative adversarial network was applied in combination with class-conditioned convolution layers. With this novel task, VR images and videos will be easily created even with a smartphone camera.

18.MS$^2$L: Multi-Task Self-Supervised Learning for Skeleton Based Action Recognition ⬇️

In this paper, we address self-supervised representation learning from human skeletons for action recognition. Previous methods, which usually learn feature presentations from a single reconstruction task, may come across the overfitting problem, and the features are not generalizable for action recognition. Instead, we propose to integrate multiple tasks to learn more general representations in a self-supervised manner. To realize this goal, we integrate motion prediction, jigsaw puzzle recognition, and contrastive learning to learn skeleton features from different aspects. Skeleton dynamics can be modeled through motion prediction by predicting the future sequence. And temporal patterns, which are critical for action recognition, are learned through solving jigsaw puzzles. We further regularize the feature space by contrastive learning. Besides, we explore different training strategies to utilize the knowledge from self-supervised tasks for action recognition. We evaluate our multi-task self-supervised learning approach with action classifiers trained under different configurations, including unsupervised, semi-supervised and fully-supervised settings. Our experiments on the NW-UCLA, NTU RGB+D, and PKUMMD datasets show remarkable performance for action recognition, demonstrating the superiority of our method in learning more discriminative and general features.

19.High-Fidelity 3D Digital Human Creation from RGB-D Selfies ⬇️

We present a fully automatic system that can produce high-fidelity, photo-realistic 3D digital human characters with a consumer RGB-D selfie camera. The system only needs the user to take a short selfie RGB-D video while rotating his/her head, and can produce a high quality reconstruction in less than 30 seconds. Our main contribution is a new facial geometry modeling and reflectance synthesis procedure that significantly improves the state-of-the-art. Specifically, given the input video a two-stage frame selection algorithm is first employed to select a few high-quality frames for reconstruction. A novel, differentiable renderer based 3D Morphable Model (3DMM) fitting method is then applied to recover facial geometries from multiview RGB-D data, which takes advantages of extensive data generation and perturbation. Our 3DMM has much larger expressive capacities than conventional 3DMM, allowing us to recover more accurate facial geometry using merely linear bases. For reflectance synthesis, we present a hybrid approach that combines parametric fitting and CNNs to synthesize high-resolution albedo/normal maps with realistic hair/pore/wrinkle details. Results show that our system can produce faithful 3D characters with extremely realistic details. Code and the constructed 3DMM is publicly available.

20.Learning Selective Mutual Attention and Contrast for RGB-D Saliency Detection ⬇️

How to effectively fuse cross-modal information is the key problem for RGB-D salient object detection. Early fusion and the result fusion schemes fuse RGB and depth information at the input and output stages, respectively, hence incur the problem of distribution gap or information loss. Many models use the feature fusion strategy but are limited by the low-order point-to-point fusion methods. In this paper, we propose a novel mutual attention model by fusing attention and contexts from different modalities. We use the non-local attention of one modality to propagate long-range contextual dependencies for the other modality, thus leveraging complementary attention cues to perform high-order and trilinear cross-modal interaction. We also propose to induce contrast inference from the mutual attention and obtain a unified model. Considering low-quality depth data may detriment the model performance, we further propose selective attention to reweight the added depth cues. We embed the proposed modules in a two-stream CNN for RGB-D SOD. Experimental results have demonstrated the effectiveness of our proposed model. Moreover, we also construct a new challenging large-scale RGB-D SOD dataset with high-quality, thus can both promote the training and evaluation of deep models.

21.Unsupervised Semantic Aggregation and Deformable Template Matching for Semi-Supervised Learning ⬇️

Unlabeled data learning has attracted considerable attention recently. However, it is still elusive to extract the expected high-level semantic feature with mere unsupervised learning. In the meantime, semi-supervised learning (SSL) demonstrates a promising future in leveraging few samples. In this paper, we combine both to propose an Unsupervised Semantic Aggregation and Deformable Template Matching (USADTM) framework for SSL, which strives to improve the classification performance with few labeled data and then reduce the cost in data annotating. Specifically, unsupervised semantic aggregation based on Triplet Mutual Information (T-MI) loss is explored to generate semantic labels for unlabeled data. Then the semantic labels are aligned to the actual class by the supervision of labeled data. Furthermore, a feature pool that stores the labeled samples is dynamically updated to assign proxy labels for unlabeled data, which are used as targets for cross-entropy minimization. Extensive experiments and analysis across four standard semi-supervised learning benchmarks validate that USADTM achieves top performance (e.g., 90.46$%$ accuracy on CIFAR-10 with 40 labels and 95.20$%$ accuracy with 250 labels). The code is released at this https URL.

22.Automatic Quantification of Settlement Damage using Deep Learning of Satellite Images ⬇️

Humanitarian disasters and political violence cause significant damage to our living space. The reparation cost to homes, infrastructure, and the ecosystem is often difficult to quantify in real-time. Real-time quantification is critical to both informing relief operations, but also planning ahead for rebuilding. Here, we use satellite images before and after major crisis around the world to train a robust baseline Residual Network (ResNet) and a disaster quantification Pyramid Scene Parsing Network (PSPNet). ResNet offers robustness to poor image quality and can identify areas of destruction with high accuracy (92%), whereas PSPNet offers contextualised quantification of built environment damage with good accuracy (84%). As there are multiple damage dimensions to consider (e.g. economic loss and fatalities), we fit a multi-linear regression model to quantify the overall damage. To validate our combined system of deep learning and regression modeling, we successfully match our prediction to the ongoing recovery in the 2020 Beirut port explosion. These innovations provide a better quantification of overall disaster magnitude and inform intelligent humanitarian systems of unfolding disasters.

23.Implicit Subspace Prior Learning for Dual-Blind Face Restoration ⬇️

Face restoration is an inherently ill-posed problem, where additional prior constraints are typically considered crucial for mitigating such pathology. However, real-world image prior are often hard to simulate with precise mathematical models, which inevitably limits the performance and generalization ability of existing prior-regularized restoration methods. In this paper, we study the problem of face restoration under a more practical ``dual blind'' setting, i.e., without prior assumptions or hand-crafted regularization terms on the degradation profile or image contents.
To this end, a novel implicit subspace prior learning (ISPL) framework is proposed as a generic solution to dual-blind face restoration, with two key elements: 1) an implicit formulation to circumvent the ill-defined restoration mapping and 2) a subspace prior decomposition and fusion mechanism to dynamically handle inputs at varying degradation levels with consistent high-quality restoration results.
Experimental results demonstrate significant perception-distortion improvement of ISPL against existing state-of-the-art methods for a variety of restoration subtasks, including a 3.69db PSNR and 45.8% FID gain against ESRGAN, the 2018 NTIRE SR challenge winner. Overall, we prove that it is possible to capture and utilize prior knowledge without explicitly formulating it, which will help inspire new research paradigms towards low-level vision tasks.

24.Scene Gated Social Graph: Pedestrian Trajectory Prediction Based on Dynamic Social Graphs and Scene Constraints ⬇️

Pedestrian trajectory prediction is valuable for understanding human motion behaviors and it is challenging because of the social influence from other pedestrians, the scene constraints and the multimodal possibilities of predicted trajectories. Most existing methods only focus on two of the above three key elements. In order to jointly consider all these elements, we propose a novel trajectory prediction method named Scene Gated Social Graph (SGSG). In the proposed SGSG, dynamic graphs are used to describe the social relationship among pedestrians. The social and scene influences are taken into account through the scene gated social graph features which combine the encoded social graph features and semantic scene features. In addition, a VAE module is incorporated to learn the scene gated social feature and sample latent variables for generating multiple trajectories that are socially and environmentally acceptable. We compare our SGSG against twenty state-of-the-art pedestrian trajectory prediction methods and the results show that the proposed method achieves superior performance on two widely used trajectory prediction benchmarks.

25.BiPointNet: Binary Neural Network for Point Clouds ⬇️

To alleviate the resource constraint for real-time point clouds applications that run on edge devices, we present BiPointNet, the first model binarization approach for efficient deep learning on point clouds. In this work, we discover that the immense performance drop of binarized models for point clouds is caused by two main challenges: aggregation-induced feature homogenization that leads to a degradation of information entropy, and scale distortion that hinders optimization and invalidates scale-sensitive structures. With theoretical justifications and in-depth analysis, we propose Entropy-Maximizing Aggregation(EMA) to modulate the distribution before aggregation for the maximum information entropy, andLayer-wise Scale Recovery(LSR) to efficiently restore feature scales. Extensive experiments show that our BiPointNet outperforms existing binarization methods by convincing margins, at the level even comparable with the full precision counterpart. We highlight that our techniques are generic which show significant improvements on various fundamental tasks and mainstream backbones. BiPoint-Net gives an impressive 14.7 times speedup and 18.9 times storage saving on real-world resource-constrained devices.

26.Increasing the Robustness of Semantic Segmentation Models with Painting-by-Numbers ⬇️

For safety-critical applications such as autonomous driving, CNNs have to be robust with respect to unavoidable image corruptions, such as image noise. While previous works addressed the task of robust prediction in the context of full-image classification, we consider it for dense semantic segmentation. We build upon an insight from image classification that output robustness can be improved by increasing the network-bias towards object shapes. We present a new training schema that increases this shape bias. Our basic idea is to alpha-blend a portion of the RGB training images with faked images, where each class-label is given a fixed, randomly chosen color that is not likely to appear in real imagery. This forces the network to rely more strongly on shape cues. We call this data augmentation technique ``Painting-by-Numbers''. We demonstrate the effectiveness of our training schema for DeepLabv3+ with various network backbones, MobileNet-V2, ResNets, and Xception, and evaluate it on the Cityscapes dataset. With respect to our 16 different types of image corruptions and 5 different network backbones, we are in 74% better than training with clean data. For cases where we are worse than a model trained without our training schema, it is mostly only marginally worse. However, for some image corruptions such as images with noise, we see a considerable performance gain of up to 25%.

27.CC-Loss: Channel Correlation Loss For Image Classification ⬇️

The loss function is a key component in deep learning models. A commonly used loss function for classification is the cross entropy loss, which is a simple yet effective application of information theory for classification problems. Based on this loss, many other loss functions have been proposed,~\emph{e.g.}, by adding intra-class and inter-class constraints to enhance the discriminative ability of the learned features. However, these loss functions fail to consider the connections between the feature distribution and the model structure. Aiming at addressing this problem, we propose a channel correlation loss (CC-Loss) that is able to constrain the specific relations between classes and channels as well as maintain the intra-class and the inter-class separability. CC-Loss uses a channel attention module to generate channel attention of features for each sample in the training stage. Next, an Euclidean distance matrix is calculated to make the channel attention vectors associated with the same class become identical and to increase the difference between different classes. Finally, we obtain a feature embedding with good intra-class compactness and inter-class separability.Experimental results show that two different backbone models trained with the proposed CC-Loss outperform the state-of-the-art loss functions on three image classification datasets.

28.TSPNet: Hierarchical Feature Learning via Temporal Semantic Pyramid for Sign Language Translation ⬇️

Sign language translation (SLT) aims to interpret sign video sequences into text-based natural language sentences. Sign videos consist of continuous sequences of sign gestures with no clear boundaries in between. Existing SLT models usually represent sign visual features in a frame-wise manner so as to avoid needing to explicitly segmenting the videos into isolated signs. However, these methods neglect the temporal information of signs and lead to substantial ambiguity in translation. In this paper, we explore the temporal semantic structures of signvideos to learn more discriminative features. To this end, we first present a novel sign video segment representation which takes into account multiple temporal granularities, thus alleviating the need for accurate video segmentation. Taking advantage of the proposed segment representation, we develop a novel hierarchical sign video feature learning method via a temporal semantic pyramid network, called TSPNet. Specifically, TSPNet introduces an inter-scale attention to evaluate and enhance local semantic consistency of sign segments and an intra-scale attention to resolve semantic ambiguity by using non-local video context. Experiments show that our TSPNet outperforms the state-of-the-art with significant improvements on the BLEU score (from 9.58 to 13.41) and ROUGE score (from 31.80 to 34.96)on the largest commonly-used SLT dataset. Our implementation is available at this https URL.

29.Discriminative Sounding Objects Localization via Self-supervised Audiovisual Matching ⬇️

Discriminatively localizing sounding objects in cocktail-party, i.e., mixed sound scenes, is commonplace for humans, but still challenging for machines. In this paper, we propose a two-stage learning framework to perform self-supervised class-aware sounding object localization. First, we propose to learn robust object representations by aggregating the candidate sound localization results in the single source scenes. Then, class-aware object localization maps are generated in the cocktail-party scenarios by referring the pre-learned object knowledge, and the sounding objects are accordingly selected by matching audio and visual object category distributions, where the audiovisual consistency is viewed as the self-supervised signal. Experimental results in both realistic and synthesized cocktail-party videos demonstrate that our model is superior in filtering out silent objects and pointing out the location of sounding objects of different classes. Code is available at this https URL.

30.Top-DB-Net: Top DropBlock for Activation Enhancement in Person Re-Identification ⬇️

Person Re-Identification is a challenging task that aims to retrieve all instances of a query image across a system of non-overlapping cameras. Due to the various extreme changes of view, it is common that local regions that could be used to match people are suppressed, which leads to a scenario where approaches have to evaluate the similarity of images based on less informative regions. In this work, we introduce the Top-DB-Net, a method based on Top DropBlock that pushes the network to learn to focus on the scene foreground, with special emphasis on the most task-relevant regions and, at the same time, encodes low informative regions to provide high discriminability. The Top-DB-Net is composed of three streams: (i) a global stream encodes rich image information from a backbone, (ii) the Top DropBlock stream encourages the backbone to encode low informative regions with high discriminative features, and (iii) a regularization stream helps to deal with the noise created by the dropping process of the second stream, when testing the first two streams are used. Vast experiments on three challenging datasets show the capabilities of our approach against state-of-the-art methods. Qualitative results demonstrate that our method exhibits better activation maps focusing on reliable parts of the input images.

31.Reconstruction of Quantitative Susceptibility Maps from Phase of Susceptibility Weighted Imaging with Cross-Connected ψ-Net ⬇️

Quantitative Susceptibility Mapping (QSM) is a new phase-based technique for quantifying magnetic susceptibility. The existing QSM reconstruction methods generally require complicated pre-processing on high-quality phase data. In this work, we propose to explore a new value of the high-pass filtered phase data generated in susceptibility weighted imaging (SWI), and develop an end-to-end Cross-connected {\psi}-Net (C{\psi}-Net) to reconstruct QSM directly from these phase data in SWI without additional pre-processing. C{\psi}-Net adds an intermediate branch in the classical U-Net to form a {\psi}-like structure. The specially designed dilated interaction block is embedded in each level of this branch to enlarge the receptive fields for capturing more susceptibility information from a wider spatial range of phase images. Moreover, the crossed connections are utilized between branches to implement a multi-resolution feature fusion scheme, which helps C{\psi}-Net capture rich contextual information for accurate reconstruction. The experimental results on a human dataset show that C{\psi}-Net achieves superior performance in our task over other QSM reconstruction algorithms.

32.A Progressive Conditional Generative Adversarial Network for Generating Dense and Colored 3D Point Clouds ⬇️

In this paper, we introduce a novel conditional generative adversarial network that creates dense 3D point clouds, with color, for assorted classes of objects in an unsupervised manner. To overcome the difficulty of capturing intricate details at high resolutions, we propose a point transformer that progressively grows the network through the use of graph convolutions. The network is composed of a leaf output layer and an initial set of branches. Every training iteration evolves a point vector into a point cloud of increasing resolution. After a fixed number of iterations, the number of branches is increased by replicating the last branch. Experimental results show that our network is capable of learning and mimicking a 3D data distribution, and produces colored point clouds with fine details at multiple resolutions.

33.Medical Image Harmonization Using Deep Learning Based Canonical Mapping: Toward Robust and Generalizable Learning in Imaging ⬇️

Conventional and deep learning-based methods have shown great potential in the medical imaging domain, as means for deriving diagnostic, prognostic, and predictive biomarkers, and by contributing to precision medicine. However, these methods have yet to see widespread clinical adoption, in part due to limited generalization performance across various imaging devices, acquisition protocols, and patient populations. In this work, we propose a new paradigm in which data from a diverse range of acquisition conditions are "harmonized" to a common reference domain, where accurate model learning and prediction can take place. By learning an unsupervised image to image canonical mapping from diverse datasets to a reference domain using generative deep learning models, we aim to reduce confounding data variation while preserving semantic information, thereby rendering the learning task easier in the reference domain. We test this approach on two example problems, namely MRI-based brain age prediction and classification of schizophrenia, leveraging pooled cohorts of neuroimaging MRI data spanning 9 sites and 9701 subjects. Our results indicate a substantial improvement in these tasks in out-of-sample data, even when training is restricted to a single site.

34.MoCo Pretraining Improves Representation and Transferability of Chest X-ray Models ⬇️

Self-supervised approaches such as Momentum Contrast (MoCo) can leverage unlabeled data to produce pretrained models for subsequent fine-tuning on labeled data. While MoCo has demonstrated promising results on natural image classification tasks, its application to medical imaging tasks like chest X-ray interpretation has been limited. Chest X-ray interpretation is fundamentally different from natural image classification in ways that may limit the applicability of self-supervised approaches. In this work, we investigate whether MoCo-pretraining leads to better representations or initializations for chest X-ray interpretation. We conduct MoCo-pretraining on CheXpert, a large labeled dataset of X-rays, followed by supervised fine-tuning experiments on the pleural effusion task. Using 0.1% of labeled training data, we find that a linear model trained on MoCo-pretrained representations outperforms one trained on representations without MoCo-pretraining by an AUC of 0.096 (95% CI 0.061, 0.130), indicating that MoCo-pretrained representations are of higher quality. Furthermore, a model fine-tuned end-to-end with MoCo-pretraining outperforms its non-MoCo-pretrained counterpart by an AUC of 0.037 (95% CI 0.015, 0.062) with the 0.1% label fraction. These AUC improvements are observed for all label fractions for both the linear model and an end-to-end fine-tuned model with the greater improvements for smaller label fractions. Finally, we observe similar results on a small, target chest X-ray dataset (Shenzhen dataset for tuberculosis) with MoCo-pretraining done on the source dataset (CheXpert), which suggests that pretraining on unlabeled X-rays can provide transfer learning benefits for a target task. Our study demonstrates that MoCo-pretraining provides high-quality representations and transferable initializations for chest X-ray interpretation.

35.Identifying Melanoma Images using EfficientNet Ensemble: Winning Solution to the SIIM-ISIC Melanoma Classification Challenge ⬇️

We present our winning solution to the SIIM-ISIC Melanoma Classification Challenge. It is an ensemble of convolutions neural network (CNN) models with different backbones and input sizes, most of which are image-only models while a few of them used image-level and patient-level metadata. The keys to our winning are: (1) stable validation scheme (2) good choice of model target (3) carefully tuned pipeline and (4) ensembling with very diverse models. The winning submission scored 0.9600 AUC on cross validation and 0.9490 AUC on private leaderboard.

36.Google Landmark Recognition 2020 Competition Third Place Solution ⬇️

We present our third place solution to the Google Landmark Recognition 2020 competition. It is an ensemble of global features only Sub-center ArcFace models. We introduce dynamic margins for ArcFace loss, a family of tune-able margin functions of class size, designed to deal with the extreme imbalance in GLDv2 dataset. Progressive finetuning and careful postprocessing are also key to the solution. Our two submissions scored 0.6344 and 0.6289 on private leaderboard, both ranking third place out of 736 teams.

37.Self-attention aggregation network for video face representation and recognition ⬇️

Models based on self-attention mechanisms have been successful in analyzing temporal data and have been widely used in the natural language domain. We propose a new model architecture for video face representation and recognition based on a self-attention mechanism. Our approach could be used for video with single and multiple identities. To the best of our knowledge, no one has explored the aggregation approaches that consider the video with multiple identities. The proposed approach utilizes existing models to get the face representation for each video frame, e.g., ArcFace and MobileFaceNet, and the aggregation module produces the aggregated face representation vector for video by taking into consideration the order of frames and their quality scores. We demonstrate empirical results on a public dataset for video face recognition called IJB-C to indicate that the self-attention aggregation network (SAAN) outperforms naive average pooling. Moreover, we introduce a new multi-identity video dataset based on the publicly available UMDFaces dataset and collected GIFs from Giphy. We show that SAAN is capable of producing a compact face representation for both single and multiple identities in a video. The dataset and source code will be publicly available.

38.Resolution Dependant GAN Interpolation for Controllable Image Synthesis Between Domains ⬇️

GANs can generate photo-realistic images from the domain of their training data. However, those wanting to use them for creative purposes often want to generate imagery from a truly novel domain, a task which GANs are inherently unable to do. It is also desirable to have a level of control so that there is a degree of artistic direction rather than purely curation of random results. Here we present a method for interpolating between generative models of the StyleGAN architecture in a resolution dependant manner. This allows us to generate images from an entirely novel domain and do this with a degree of control over the nature of the output.

39.Revising FUNSD dataset for key-value detection in document images ⬇️

FUNSD is one of the limited publicly available datasets for information extraction from document im-ages. The information in the FUNSD dataset is defined by text areas of four categories ("key", "value", "header", "other", and "background") and connectivity between areas as key-value relations. In-specting FUNSD, we found several inconsistency in labeling, which impeded its applicability to thekey-value extraction problem. In this report, we described some labeling issues in FUNSD and therevision we made to the dataset. We also reported our implementation of for key-value detection onFUNSD using a UNet model as baseline results and an improved UNet model with Channel-InvariantDeformable Convolution.

40.H2O-Net: Self-Supervised Flood Segmentation via Adversarial Domain Adaptation and Label Refinement ⬇️

Accurate flood detection in near real time via high resolution, high latency satellite imagery is essential to prevent loss of lives by providing quick and actionable information. Instruments and sensors useful for flood detection are only available in low resolution, low latency satellites with region re-visit periods of up to 16 days, making flood alerting systems that use such satellites unreliable. This work presents H2O-Network, a self supervised deep learning method to segment floods from satellites and aerial imagery by bridging domain gap between low and high latency satellite and coarse-to-fine label refinement. H2O-Net learns to synthesize signals highly correlative with water presence as a domain adaptation step for semantic segmentation in high resolution satellite imagery. Our work also proposes a self-supervision mechanism, which does not require any hand annotation, used during training to generate high quality ground truth data. We demonstrate that H2O-Net outperforms the state-of-the-art semantic segmentation methods on satellite imagery by 10% and 12% pixel accuracy and mIoU respectively for the task of flood segmentation. We emphasize the generalizability of our model by transferring model weights trained on satellite imagery to drone imagery, a highly different sensor and domain.

41.PI-Net: Pose Interacting Network for Multi-Person Monocular 3D Pose Estimation ⬇️

Recent literature addressed the monocular 3D pose estimation task very satisfactorily. In these studies, different persons are usually treated as independent pose instances to estimate. However, in many every-day situations, people are interacting, and the pose of an individual depends on the pose of his/her interactees. In this paper, we investigate how to exploit this dependency to enhance current - and possibly future - deep networks for 3D monocular pose estimation. Our pose interacting network, or PI-Net, inputs the initial pose estimates of a variable number of interactees into a recurrent architecture used to refine the pose of the person-of-interest. Evaluating such a method is challenging due to the limited availability of public annotated multi-person 3D human pose datasets. We demonstrate the effectiveness of our method in the MuPoTS dataset, setting the new state-of-the-art on it. Qualitative results on other multi-person datasets (for which 3D pose ground-truth is not available) showcase the proposed PI-Net. PI-Net is implemented in PyTorch and the code will be made available upon acceptance of the paper.

42.Glance and Focus: a Dynamic Approach to Reducing Spatial Redundancy in Image Classification ⬇️

The accuracy of deep convolutional neural networks (CNNs) generally improves when fueled with high resolution images. However, this often comes at a high computational cost and high memory footprint. Inspired by the fact that not all regions in an image are task-relevant, we propose a novel framework that performs efficient image classification by processing a sequence of relatively small inputs, which are strategically selected from the original image with reinforcement learning. Such a dynamic decision process naturally facilitates adaptive inference at test time, i.e., it can be terminated once the model is sufficiently confident about its prediction and thus avoids further redundant computation. Notably, our framework is general and flexible as it is compatible with most of the state-of-the-art light-weighted CNNs (such as MobileNets, EfficientNets and RegNets), which can be conveniently deployed as the backbone feature extractor. Experiments on ImageNet show that our method consistently improves the computational efficiency of a wide variety of deep models. For example, it further reduces the average latency of the highly efficient MobileNet-V3 on an iPhone XS Max by 20% without sacrificing accuracy. Code and pre-trained models are available at this https URL.

43.IF-Defense: 3D Adversarial Point Cloud Defense via Implicit Function based Restoration ⬇️

Point cloud is an important 3D data representation widely used in many essential applications. Leveraging deep neural networks, recent works have shown great success in processing 3D point clouds. However, those deep neural networks are vulnerable to various 3D adversarial attacks, which can be summarized as two primary types: point perturbation that affects local point distribution, and surface distortion that causes dramatic changes in geometry. In this paper, we propose a novel 3D adversarial point cloud defense method leveraging implicit function based restoration (IF-Defense) to address both the aforementioned attacks. It is composed of two steps: 1) it predicts an implicit function that captures the clean shape through a surface recovery module, and 2) restores a clean and complete point cloud via minimizing the difference between the attacked point cloud and the predicted implicit function under geometry- and distribution- aware constraints. Our experimental results show that IF-Defense achieves the state-of-the-art defense performance against all existing adversarial attacks on PointNet, PointNet++, DGCNN and PointConv. Comparing with previous methods, IF-Defense presents 20.02% improvement in classification accuracy against salient point dropping attack and 16.29% against LG-GAN attack on PointNet.

44.Boosting Continuous Sign Language Recognition via Cross Modality Augmentation ⬇️

Continuous sign language recognition (SLR) deals with unaligned video-text pair and uses the word error rate (WER), i.e., edit distance, as the main evaluation metric. Since it is not differentiable, we usually instead optimize the learning model with the connectionist temporal classification (CTC) objective loss, which maximizes the posterior probability over the sequential alignment. Due to the optimization gap, the predicted sentence with the highest decoding probability may not be the best choice under the WER metric. To tackle this issue, we propose a novel architecture with cross modality augmentation. Specifically, we first augment cross-modal data by simulating the calculation procedure of WER, i.e., substitution, deletion and insertion on both text label and its corresponding video. With these real and generated pseudo video-text pairs, we propose multiple loss terms to minimize the cross modality distance between the video and ground truth label, and make the network distinguish the difference between real and pseudo modalities. The proposed framework can be easily extended to other existing CTC based continuous SLR architectures. Extensive experiments on two continuous SLR benchmarks, i.e., RWTH-PHOENIX-Weather and CSL, validate the effectiveness of our proposed method.

45.Infrared target tracking based on proximal robust principal component analysis method ⬇️

Infrared target tracking plays an important role in both civil and military fields. The main challenges in designing a robust and high-precision tracker for infrared sequences include overlap, occlusion and appearance change. To this end, this paper proposes an infrared target tracker based on proximal robust principal component analysis method. Firstly, the observation matrix is decomposed into a sparse occlusion matrix and a low-rank target matrix, and the constraint optimization is carried out with an approaching proximal norm which is better than L1-norm. To solve this convex optimization problem, Alternating Direction Method of Multipliers (ADMM) is employed to estimate the variables alternately. Finally, the framework of particle filter with model update strategy is exploited to locate the target. Through a series of experiments on real infrared target sequences, the effectiveness and robustness of our algorithm are proved.

46.Shape-aware Generative Adversarial Networks for Attribute Transfer ⬇️

Generative adversarial networks (GANs) have been successfully applied to transfer visual attributes in many domains, including that of human face images. This success is partly attributable to the facts that human faces have similar shapes and the positions of eyes, noses, and mouths are fixed among different people. Attribute transfer is more challenging when the source and target domain share different shapes. In this paper, we introduce a shape-aware GAN model that is able to preserve shape when transferring attributes, and propose its application to some real-world domains. Compared to other state-of-art GANs-based image-to-image translation models, the model we propose is able to generate more visually appealing results while maintaining the quality of results from transfer learning.

47.Partial FC: Training 10 Million Identities on a Single Machine ⬇️

Face recognition has been an active and vital topic among computer vision community for a long time. Previous researches mainly focus on loss functions used for facial feature extraction network, among which the improvements of softmax-based loss functions greatly promote the performance of face recognition. However, the contradiction between the drastically increasing number of face identities and the shortage of GPU memories is gradually becoming irreconcilable. In this paper, we thoroughly analyze the optimization goal of softmax-based loss functions and the difficulty of training massive identities. We find that the importance of negative classes in softmax function in face representation learning is not as high as we previously thought. The experiment demonstrates no loss of accuracy when training with only 10% randomly sampled classes for the softmax-based loss functions, compared with training with full classes using state-of-the-art models on mainstream benchmarks. We also implement a very efficient distributed sampling algorithm, taking into account model accuracy and training efficiency, which uses only eight NVIDIA RTX2080Ti to complete classification tasks with tens of millions of identities. The code of this paper has been made available this https URL.

48.GuCNet: A Guided Clustering-based Network for Improved Classification ⬇️

We deal with the problem of semantic classification of challenging and highly-cluttered dataset. We present a novel, and yet a very simple classification technique by leveraging the ease of classifiability of any existing well separable dataset for guidance. Since the guide dataset which may or may not have any semantic relationship with the experimental dataset, forms well separable clusters in the feature set, the proposed network tries to embed class-wise features of the challenging dataset to those distinct clusters of the guide set, making them more separable. Depending on the availability, we propose two types of guide sets: one using texture (image) guides and another using prototype vectors representing cluster centers. Experimental results obtained on the challenging benchmark RSSCN, LSUN, and TU-Berlin datasets establish the efficacy of the proposed method as we outperform the existing state-of-the-art techniques by a considerable margin.

49.Generalized Few-Shot Semantic Segmentation ⬇️

Training semantic segmentation models requires a large amount of finely annotated data, making it hard to quickly adapt to novel classes not satisfying this condition. Few-Shot Segmentation (FS-Seg) tackles this problem with many constraints. In this paper, we introduce a new benchmark, called Generalized Few-Shot Semantic Segmentation (GFS-Seg), to analyze the generalization ability of segmentation models to simultaneously recognize novel categories with very few examples as well as base categories with sufficient examples. Previous state-of-the-art FS-Seg methods fall short in GFS-Seg and the performance discrepancy mainly comes from the constrained training setting of FS-Seg. To make GFS-Seg tractable, we set up a GFS-Seg baseline that achieves decent performance without structural change on the original model. Then, as context is the key for boosting performance on semantic segmentation, we propose the Context-Aware Prototype Learning (CAPL) that significantly improves performance by leveraging the contextual information to update class prototypes with aligned features. Extensive experiments on Pascal-VOC and COCO manifest the effectiveness of CAPL, and CAPL also generalizes well to FS-Seg.

50.Constructing a Visual Relationship Authenticity Dataset ⬇️

A visual relationship denotes a relationship between two objects in an image, which can be represented as a triplet of (subject; predicate; object). Visual relationship detection is crucial for scene understanding in images. Existing visual relationship detection datasets only contain true relationships that correctly describe the content in an image. However, distinguishing false visual relationships from true ones is also crucial for image understanding and grounded natural language processing. In this paper, we construct a visual relationship authenticity dataset, where both true and false relationships among all objects appeared in the captions in the Flickr30k entities image caption dataset are annotated. The dataset is available at this https URL. We hope that this dataset can promote the study on both vision and language understanding.

51.MammoGANesis: Controlled Generation of High-Resolution Mammograms for Radiology Education ⬇️

During their formative years, radiology trainees are required to interpret hundreds of mammograms per month, with the objective of becoming apt at discerning the subtle patterns differentiating benign from malignant lesions. Unfortunately, medico-legal and technical hurdles make it difficult to access and query medical images for training.
In this paper we train a generative adversarial network (GAN) to synthesize 512 x 512 high-resolution mammograms. The resulting model leads to the unsupervised separation of high-level features (e.g. the standard mammography views and the nature of the breast lesions), with stochastic variation in the generated images (e.g. breast adipose tissue, calcification), enabling user-controlled global and local attribute-editing of the synthesized images.
We demonstrate the model's ability to generate anatomically and medically relevant mammograms by achieving an average AUC of 0.54 in a double-blind study on four expert mammography radiologists to distinguish between generated and real images, ascribing to the high visual quality of the synthesized and edited mammograms, and to their potential use in advancing and facilitating medical education.

52.SDE-AWB: a Generic Solution for 2nd International Illumination Estimation Challenge ⬇️

We propose a neural network-based solution for three different tracks of 2nd International Illumination Estimation Challenge (this http URL). Our method is built on pre-trained Squeeze-Net backbone, differential 2D chroma histogram layer and a shallow MLP utilizing Exif information. By combining semantic feature, color feature and Exif metadata, the resulting method -- SDE-AWB -- obtains 1st place in both indoor and two-illuminant tracks and 2nd place in general track.

53.SDMTL: Semi-Decoupled Multi-grained Trajectory Learning for 3D human motion prediction ⬇️

Predicting future human motion is critical for intelligent robots to interact with humans in the real world, and human motion has the nature of multi-granularity. However, most of the existing work either implicitly modeled multi-granularity information via fixed modes or focused on modeling a single granularity, making it hard to well capture this nature for accurate predictions. In contrast, we propose a novel end-to-end network, Semi-Decoupled Multi-grained Trajectory Learning network (SDMTL), to predict future poses, which not only flexibly captures rich multi-grained trajectory information but also aggregates multi-granularity information for predictions. Specifically, we first introduce a Brain-inspired Semi-decoupled Motion-sensitive Encoding module (BSME), effectively capturing spatiotemporal features in a semi-decoupled manner. Then, we capture the temporal dynamics of motion trajectory at multi-granularity, including fine granularity and coarse granularity. We learn multi-grained trajectory information using BSMEs hierarchically and further capture the information of temporal evolutional directions at each granularity by gathering the outputs of BSMEs at each granularity and applying temporal convolutions along the motion trajectory. Next, the captured motion dynamics can be further enhanced by aggregating the information of multi-granularity with a weighted summation scheme. Finally, experimental results on two benchmarks, including Human3.6M and CMU-Mocap, show that our method achieves state-of-the-art performance, demonstrating the effectiveness of our proposed method. The code will be available if the paper is accepted.

54.Segmenting Epipolar Line ⬇️

Identifying feature correspondence between two images is a fundamental procedure in three-dimensional computer vision. Usually the feature search space is confined by the epipolar line. Using the cheirality constraint, this paper finds that the feature search space can be restrained to one of two or three segments of the epipolar line that are defined by the epipole and a so-called virtual infinity point.

55.Anomaly Detection based on Zero-Shot Outlier Synthesis and Hierarchical Feature Distillation ⬇️

Anomaly detection suffers from unbalanced data since anomalies are quite rare. Synthetically generated anomalies are a solution to such ill or not fully defined data. However, synthesis requires an expressive representation to guarantee the quality of the generated data. In this paper, we propose a two-level hierarchical latent space representation that distills inliers' feature-descriptors (through autoencoders) into more robust representations based on a variational family of distributions (through a variational autoencoder) for zero-shot anomaly generation. From the learned latent distributions, we select those that lie on the outskirts of the training data as synthetic-outlier generators. And, we synthesize from them, i.e., generate negative samples without seen them before, to train binary classifiers. We found that the use of the proposed hierarchical structure for feature distillation and fusion creates robust and general representations that allow us to synthesize pseudo outlier samples. And in turn, train robust binary classifiers for true outlier detection (without the need for actual outliers during training). We demonstrate the performance of our proposal on several benchmarks for anomaly detection.

56.Diagnosing and Preventing Instabilities in Recurrent Video Processing ⬇️

Recurrent models are becoming a popular choice for video enhancement tasks such as video denoising. In this work, we focus on their stability as dynamical systems and show that they tend to fail catastrophically at inference time on long video sequences. To address this issue, we (1) introduce a diagnostic tool which produces adversarial input sequences optimized to trigger instabilities and that can be interpreted as visualizations of spatio-temporal receptive fields, and (2) propose two approaches to enforce the stability of a model: constraining the spectral norm or constraining the stable rank of its convolutional layers. We then introduce Stable Rank Normalization of the Layers (SRNL), a new algorithm that enforces these constraints, and verify experimentally that it successfully results in stable recurrent video processing.

57.Hybrid Sequence to Sequence Model for Video Object Segmentation ⬇️

One-shot Video Object Segmentation (VOS) is the task of pixel-wise tracking an object of interest within a video sequence, where the segmentation mask of the first frame is given at inference time. In recent years, Recurrent Neural Networks (RNNs) have been widely used for VOS tasks, but they often suffer from limitations such as drift and error propagation. In this work, we study an RNN-based architecture and address some of these issues by proposing a hybrid sequence-to-sequence architecture named HS2S, utilizing a hybrid mask propagation strategy that allows incorporating the information obtained from correspondence matching. Our experiments show that augmenting the RNN with correspondence matching is a highly effective solution to reduce the drift problem. The additional information helps the model to predict more accurate masks and makes it robust against error propagation. We evaluate our HS2S model on the DAVIS2017 dataset as well as Youtube-VOS. On the latter, we achieve an improvement of 11.2pp in the overall segmentation accuracy over RNN-based state-of-the-art methods in VOS. We analyze our model's behavior in challenging cases such as occlusion and long sequences and show that our hybrid architecture significantly enhances the segmentation quality in these difficult scenarios.

58.Meta-Aggregating Networks for Class-Incremental Learning ⬇️

Class-Incremental Learning (CIL) aims to learn a classification model with the number of classes increasing phase-by-phase. The inherent problem in CIL is the stability-plasticity dilemma between the learning of old and new classes, i.e., high-plasticity models easily forget old classes but high-stability models are weak to learn new classes. We alleviate this issue by proposing a novel network architecture called Meta-Aggregating Networks (MANets) in which we explicitly build two residual blocks at each residual level (taking ResNet as the baseline architecture): a stable block and a plastic block. We aggregate the output feature maps from these two blocks and then feed the results to the next-level blocks. We meta-learn the aggregating weights in order to dynamically optimize and balance between two types of blocks, i.e., between stability and plasticity. We conduct extensive experiments on three CIL benchmarks: CIFAR-100, ImageNet-Subset, and ImageNet, and show that many existing CIL methods can be straightforwardly incorporated on the architecture of MANets to boost their performance.

59.Boosted EfficientNet: Detection of Lymph Node Metastases in Breast Cancer Using Convolutional Neural Network ⬇️

In recent years, advances in the development of whole-slide images have laid a foundation for the utilization of digital images in pathology. With the assistance of computer images analysis that automatically identifies tissue or cell types, they have greatly improved the histopathologic interpretation and diagnosis accuracy. In this paper, the Convolutional Neutral Network (CNN) has been adapted to predict and classify lymph node metastasis in breast cancer. Unlike traditional image cropping methods that are only suitable for large resolution images, we propose a novel data augmentation method named Random Center Cropping (RCC) to facilitate small resolution images. RCC enriches the datasets while retaining the image resolution and the center area of images. In addition, we reduce the downsampling scale of the network to further facilitate small resolution images better. Moreover, Attention and Feature Fusion (FF) mechanisms are employed to improve the semantic information of images. Experiments demonstrate that our methods boost performances of basic CNN architectures. And the best-performed method achieves an accuracy of 97.96% and an AUC of 99.68% on RPCam datasets, respectively.

60.An Encoder-Decoder CNN for Hair Removal in Dermoscopic Images ⬇️

The process of removing occluding hair has a relevant role in the early and accurate diagnosis of skin cancer. It consists of detecting hairs and restore the texture below them, which is sporadically occluded. In this work, we present a model based on convolutional neural networks for hair removal in dermoscopic images. During the network's training, we use a combined loss function to improve the restoration ability of the proposed model. In order to train the CNN and to quantitatively validate their performance, we simulate the presence of skin hair in hairless images extracted from publicly known datasets such as the PH2, dermquest, dermis, EDRA2002, and the ISIC Data Archive. As far as we know, there is no other hair removal method based on deep learning. Thus, we compare our results with six state-of-the-art algorithms based on traditional computer vision techniques by means of similarity measures that compare the reference hairless image and the one with hair simulated. Finally, a statistical test is used to compare the methods. Both qualitative and quantitative results demonstrate the effectiveness of our network.

61.Light Field Salient Object Detection: A Review and Benchmark ⬇️

Salient object detection (SOD) is a long-standing research topic in computer vision and has drawn an increasing amount of research interest in the past decade. This paper provides the first comprehensive review and benchmark for SOD on light field, which has long been lacking in the saliency community. Firstly, we introduce the preliminary knowledge of light field including theory and data forms, and then review existing studies on light field SOD, covering ten traditional models, six deep learning-based models, one comparative study, and one brief review. Existing datasets for light field SOD are summarized with detailed information and statistical analysis. Secondly, we benchmark seven representative light field SOD models together with several cutting-edge RGB-D SOD models on four widely used light field datasets, from which insightful discussions and analyses including the comparison between light field SOD and RGB-D SOD models are achieved. Besides, due to the inconsistency of datasets in their current forms, we further generate complete data and supplement focal stacks, depth maps and multi-view images for the inconsistent datasets, making them consistent and unified. Our supplemented data makes a universal benchmark possible. Lastly, because light field SOD is a quite special problem attributed to its diverse data representations and high dependency on acquisition hardware, making it differ greatly from other saliency detection tasks, we provide nine hints into the challenges and future directions, and outline several open issues. We hope our review and benchmarking could serve as a catalyst to advance research in this field. All the materials including collected models, datasets, benchmarking results, and supplemented light field datasets will be publicly available at our project site.

62.HCNet: Hierarchical Context Network for Semantic Segmentation ⬇️

Global context information is vital in visual understanding problems, especially in pixel-level semantic segmentation. The mainstream methods adopt the self-attention mechanism to model global context information. However, pixels belonging to different classes usually have weak feature correlation. Modeling the global pixel-level correlation matrix indiscriminately is extremely redundant in the self-attention mechanism. In order to solve the above problem, we propose a hierarchical context network to differentially model homogeneous pixels with strong correlations and heterogeneous pixels with weak correlations. Specifically, we first propose a multi-scale guided pre-segmentation module to divide the entire feature map into different classed-based homogeneous regions. Within each homogeneous region, we design the pixel context module to capture pixel-level correlations. Subsequently, different from the self-attention mechanism that still models weak heterogeneous correlations in a dense pixel-level manner, the region context module is proposed to model sparse region-level dependencies using a unified representation of each region. Through aggregating fine-grained pixel context features and coarse-grained region context features, our proposed network can not only hierarchically model global context information but also harvest multi-granularity representations to more robustly identify multi-scale objects. We evaluate our approach on Cityscapes and the ISPRS Vaihingen dataset. Without Bells or Whistles, our approach realizes a mean IoU of 82.8% and overall accuracy of 91.4% on Cityscapes and ISPRS Vaihingen test set, achieving state-of-the-art results.

63.Selective Information Passing for MR/CT Image Segmentation ⬇️

Automated medical image segmentation plays an important role in many clinical applications, which however is a very challenging task, due to complex background texture, lack of clear boundary and significant shape and texture variation between images. Many researchers proposed an encoder-decoder architecture with skip connections to combine low-level feature maps from the encoder path with high-level feature maps from the decoder path for automatically segmenting medical images. The skip connections have been shown to be effective in recovering fine-grained details of the target objects and may facilitate the gradient back-propagation. However, not all the feature maps transmitted by those connections contribute positively to the network performance. In this paper, to adaptively select useful information to pass through those skip connections, we propose a novel 3D network with self-supervised function, named selective information passing network (SIP-Net). We evaluate our proposed model on the MICCAI Prostate MR Image Segmentation 2012 Grant Challenge dataset, TCIA Pancreas CT-82 and MICCAI 2017 Liver Tumor Segmentation (LiTS) Challenge dataset. The experimental results across these data sets show that our model achieved improved segmentation results and outperformed other state-of-the-art methods. The source code of this work is available at this https URL.

64.Interpretable Neural Computation for Real-World Compositional Visual Question Answering ⬇️

There are two main lines of research on visual question answering (VQA): compositional model with explicit multi-hop reasoning, and monolithic network with implicit reasoning in the latent feature space. The former excels in interpretability and compositionality but fails on real-world images, while the latter usually achieves better performance due to model flexibility and parameter efficiency. We aim to combine the two to build an interpretable framework for real-world compositional VQA. In our framework, images and questions are disentangled into scene graphs and programs, and a symbolic program executor runs on them with full transparency to select the attention regions, which are then iteratively passed to a visual-linguistic pre-trained encoder to predict answers. Experiments conducted on the GQA benchmark demonstrate that our framework outperforms the compositional prior arts and achieves competitive accuracy among monolithic ones. With respect to the validity, plausibility and distribution metrics, our framework surpasses others by a considerable margin.

65.Multi-path Neural Networks for On-device Multi-domain Visual Classification ⬇️

Learning multiple domains/tasks with a single model is important for improving data efficiency and lowering inference cost for numerous vision tasks, especially on resource-constrained mobile devices. However, hand-crafting a multi-domain/task model can be both tedious and challenging. This paper proposes a novel approach to automatically learn a multi-path network for multi-domain visual classification on mobile devices. The proposed multi-path network is learned from neural architecture search by applying one reinforcement learning controller for each domain to select the best path in the super-network created from a MobileNetV3-like search space. An adaptive balanced domain prioritization algorithm is proposed to balance optimizing the joint model on multiple domains simultaneously. The determined multi-path model selectively shares parameters across domains in shared nodes while keeping domain-specific parameters within non-shared nodes in individual domain paths. This approach effectively reduces the total number of parameters and FLOPS, encouraging positive knowledge transfer while mitigating negative interference across domains. Extensive evaluations on the Visual Decathlon dataset demonstrate that the proposed multi-path model achieves state-of-the-art performance in terms of accuracy, model size, and FLOPS against other approaches using MobileNetV3-like architectures. Furthermore, the proposed method improves average accuracy over learning single-domain models individually, and reduces the total number of parameters and FLOPS by 78% and 32% respectively, compared to the approach that simply bundles single-domain models for multi-domain learning.

66.Deep Active Learning for Joint Classification & Segmentation with Weak Annotator ⬇️

CNN visualization and interpretation methods, like class activation maps (CAMs), are typically used to highlight the image regions linked to the class predictions. These models allow to simultaneously classify images and yield pixel-wise localization scores, without the need for costly pixel-level annotations. However, they are prone to high false positive localization, and thus poor visualisations when processing challenging images, such as histology images for cancer grading and localization. In this paper, an active learning (AL) framework is proposed to alleviate this issue by progressively integrating pixel-wise annotation during training. Given training data with global class-level labels, our deep weakly-supervised learning (WSL) model simultaneously allows for supervised learning for classification, and active learning for segmentation of images selected for pixel-level annotation by an oracle. Unlike traditional AL methods that focus on acquisition method, we also propose leveraging the unlabeled images to improve model accuracy with less oracle-annotation. To this end, self-learning is considered where the model is used to pseudo-annotate a large number of relevant unlabeled samples, which are then integrated during the learning process with oracle-annotated samples. Our extensive experiments are conducted on complex high resolution medical and natural images from two benchmark datasets -- GlaS for colon cancer, and CUB-200-2011 for bird species. Results indicate that by using simply random acquisition, our approach can significantly outperform segmentation obtained with state-of the-art CAMs and AL methods, using an identical oracle-supervision budget. Our method provides an efficient solution to improve the regions of interest (ROI) segmentation accuracy for real-world visual recognition applications.

67.Accelerate Your CNN from Three Dimensions: A Comprehensive Pruning Framework ⬇️

To deploy a pre-trained deep CNN on resource-constrained mobile devices, neural network pruning is often used to cut down the model's computational cost. For example, filter-level pruning (reducing the model's width) or layer-level pruning (reducing the model's depth) can both save computations with some sacrifice of accuracy. Besides, reducing the resolution of input images can also reach the same goal. Most previous methods focus on reducing one or two of these dimensions (i.e., depth, width, and image resolution) for acceleration. However, excessive reduction of any single dimension will lead to unacceptable accuracy loss, and we have to prune these three dimensions comprehensively to yield the best result. In this paper, a simple yet effective pruning framework is proposed to comprehensively consider these three dimensions. Our framework falls into two steps: 1) Determining the optimal depth (d*), width (w*), and image resolution (r) for the model. 2) Pruning the model in terms of (d*, w*, r*). Specifically, at the first step, we formulate model acceleration as an optimization problem. It takes depth (d), width (w) and image resolution (r) as variables and the model's accuracy as the optimization objective. Although it is hard to determine the expression of the objective function, approximating it with polynomials is still feasible, during which several properties of the objective function are utilized to ease and speedup the fitting process. Then the optimal d*, w* and r* are attained by maximizing the objective function with Lagrange multiplier theorem and KKT conditions. Extensive experiments are done on several popular architectures and datasets. The results show that we have outperformd the state-of-the-art pruning methods. The code will be published soon.

68.Training Binary Neural Networks through Learning with Noisy Supervision ⬇️

This paper formalizes the binarization operations over neural networks from a learning perspective. In contrast to classical hand crafted rules (\eg hard thresholding) to binarize full-precision neurons, we propose to learn a mapping from full-precision neurons to the target binary ones. Each individual weight entry will not be binarized independently. Instead, they are taken as a whole to accomplish the binarization, just as they work together in generating convolution features. To help the training of the binarization mapping, the full-precision neurons after taking sign operations is regarded as some auxiliary supervision signal, which is noisy but still has valuable guidance. An unbiased estimator is therefore introduced to mitigate the influence of the supervision noise. Experimental results on benchmark datasets indicate that the proposed binarization technique attains consistent improvements over baselines.

69.CurbScan: Curb Detection and Tracking Using Multi-Sensor Fusion ⬇️

Reliable curb detection is critical for safe autonomous driving in urban contexts. Curb detection and tracking are also useful in vehicle localization and path planning. Past work utilized a 3D LiDAR sensor to determine accurate distance information and the geometric attributes of curbs. However, such an approach requires dense point cloud data and is also vulnerable to false positives from obstacles present on both road and off-road areas. In this paper, we propose an approach to detect and track curbs by fusing together data from multiple sensors: sparse LiDAR data, a mono camera and low-cost ultrasonic sensors. The detection algorithm is based on a single 3D LiDAR and a mono camera sensor used to detect candidate curb features and it effectively removes false positives arising from surrounding static and moving obstacles. The detection accuracy of the tracking algorithm is boosted by using Kalman filter-based prediction and fusion with lateral distance information from low-cost ultrasonic sensors. We next propose a line-fitting algorithm that yields robust results for curb locations. Finally, we demonstrate the practical feasibility of our solution by testing in different road environments and evaluating our implementation in a real vehicle\footnote{Demo video clips demonstrating our algorithm have been uploaded to Youtube: this https URL, this https URL.}. Our algorithm maintains over 90% accuracy within 4.5-22 meters and 0-14 meters for the KITTI dataset and our dataset respectively, and its average processing time per frame is approximately 10 ms on Intel i7 x86 and 100ms on NVIDIA Xavier board.

70.Cluster Activation Mapping with Applications to Medical Imaging ⬇️

An open question in deep clustering is how to understand what in the image is creating the cluster assignments. This visual understanding is essential to be able to trust the results of an inherently complex algorithm like deep learning, especially when the derived cluster assignments may be used to inform decision-making or create new disease sub-types. In this work, we developed novel methodology to generate CLuster Activation Mapping (CLAM) which combines an unsupervised deep clustering framework with a modification of Score-CAM, an approach for discriminative localization in the supervised setting. We evaluated our approach using a simulation study based on computed tomography scans of the lung, and applied it to 3D CT scans from a sarcoidosis population to identify new clusters of sarcoidosis based purely on CT scan presentation.

71.Predictive Modeling of Anatomy with Genetic and Clinical Data ⬇️

We present a semi-parametric generative model for predicting anatomy of a patient in subsequent scans following a single baseline image. Such predictive modeling promises to facilitate novel analyses in both voxel-level studies and longitudinal biomarker evaluation. We capture anatomical change through a combination of population-wide regression and a non-parametric model of the subject's health based on individual genetic and clinical indicators. In contrast to classical correlation and longitudinal analysis, we focus on predicting new observations from a single subject observation. We demonstrate prediction of follow-up anatomical scans in the ADNI cohort, and illustrate a novel analysis approach that compares a patient's scans to the predicted subject-specific healthy anatomical trajectory. The code is available at this https URL.

72.Robust Optimal Transport with Applications in Generative Modeling and Domain Adaptation ⬇️

Optimal Transport (OT) distances such as Wasserstein have been used in several areas such as GANs and domain adaptation. OT, however, is very sensitive to outliers (samples with large noise) in the data since in its objective function, every sample, including outliers, is weighed similarly due to the marginal constraints. To remedy this issue, robust formulations of OT with unbalanced marginal constraints have previously been proposed. However, employing these methods in deep learning problems such as GANs and domain adaptation is challenging due to the instability of their dual optimization solvers. In this paper, we resolve these issues by deriving a computationally-efficient dual form of the robust OT optimization that is amenable to modern deep learning applications. We demonstrate the effectiveness of our formulation in two applications of GANs and domain adaptation. Our approach can train state-of-the-art GAN models on noisy datasets corrupted with outlier distributions. In particular, our optimization computes weights for training samples reflecting how difficult it is for those samples to be generated in the model. In domain adaptation, our robust OT formulation leads to improved accuracy compared to the standard adversarial adaptation methods. Our code is available at this https URL.

73.Open-sourced Dataset Protection via Backdoor Watermarking ⬇️

The rapid development of deep learning has benefited from the release of some high-quality open-sourced datasets ($e.g.$, ImageNet), which allows researchers to easily verify the effectiveness of their algorithms. Almost all existing open-sourced datasets require that they can only be adopted for academic or educational purposes rather than commercial purposes, whereas there is still no good way to protect them. In this paper, we propose a backdoor embedding based dataset watermarking method to protect an open-sourced image-classification dataset by verifying whether it is used for training a third-party model. Specifically, the proposed method contains two main processes, including dataset watermarking and dataset verification. We adopt classical poisoning-based backdoor attacks ($e.g.$, BadNets) for dataset watermarking, $i.e.$, generating some poisoned samples by adding a certain trigger ($e.g.$, a local patch) onto some benign samples, labeled with a pre-defined target class. Based on the proposed backdoor-based watermarking, we use a hypothesis test guided method for dataset verification based on the posterior probability generated by the suspicious third-party model of the benign samples and their correspondingly watermarked samples ($i.e.$, images with trigger) on the target class. Experiments on some benchmark datasets are conducted, which verify the effectiveness of the proposed method.

74.Distributionally Robust Learning for Unsupervised Domain Adaptation ⬇️

We propose a distributionally robust learning (DRL) method for unsupervised domain adaptation (UDA) that scales to modern computer vision benchmarks. DRL can be naturally formulated as a competitive two-player game between a predictor and an adversary that is allowed to corrupt the labels, subject to certain constraints, and reduces to incorporating a density ratio between the source and target domains (under the standard log loss). This formulation motivates the use of two neural networks that are jointly trained - a discriminative network between the source and target domains for density-ratio estimation, in addition to the standard classification network. The use of a density ratio in DRL prevents the model from being overconfident on target inputs far away from the source domain. Thus, DRL provides conservative confidence estimation in the target domain, even when the target labels are not available. This conservatism motivates the use of DRL in self-training for sample selection, and we term the approach distributionally robust self-training (DRST). In our experiments, DRST generates more calibrated probabilities and achieves state-of-the-art self-training accuracy on benchmark datasets. We demonstrate that DRST captures shape features more effectively, and reduces the extent of distributional shift during self-training.

75.Video Quality Enhancement Using Deep Learning-Based Prediction Models for Quantized DCT Coefficients in MPEG I-frames ⬇️

Recent works have successfully applied some types of Convolutional Neural Networks (CNNs) to reduce the noticeable distortion resulting from the lossy JPEG/MPEG compression technique. Most of them are built upon the processing made on the spatial domain. In this work, we propose a MPEG video decoder that is purely based on the frequency-to-frequency domain: it reads the quantized DCT coefficients received from a low-quality I-frames bitstream and, using a deep learning-based model, predicts the missing coefficients in order to recompose the same frames with enhanced quality. In experiments with a video dataset, our best model was able to improve from frames with quantized DCT coefficients corresponding to a Quality Factor (QF) of 10 to enhanced quality frames with QF slightly near to 20.

76.Explaining Clinical Decision Support Systems in Medical Imaging using Cycle-Consistent Activation Maximization ⬇️

Clinical decision support using deep neural networks has become a topic of steadily growing interest. While recent work has repeatedly demonstrated that deep learning offers major advantages for medical image classification over traditional methods, clinicians are often hesitant to adopt the technology because its underlying decision-making process is considered to be intransparent and difficult to comprehend. In recent years, this has been addressed by a variety of approaches that have successfully contributed to providing deeper insight. Most notably, additive feature attribution methods are able to propagate decisions back into the input space by creating a saliency map which allows the practitioner to "see what the network sees." However, the quality of the generated maps can become poor and the images noisy if only limited data is available - a typical scenario in clinical contexts. We propose a novel decision explanation scheme based on CycleGAN activation maximization which generates high-quality visualizations of classifier decisions even in smaller data sets. We conducted a user study in which these visualizations significantly outperformed existing methods on the LIDC dataset for lung lesion malignancy classification. With our approach we make a significant contribution to a better understanding of clinical decision support systems based on deep neural networks and thus aim to foster overall clinical acceptance.

77.A Unified Framework for Generic, Query-Focused, Privacy Preserving and Update Summarization using Submodular Information Measures ⬇️

We study submodular information measures as a rich framework for generic, query-focused, privacy sensitive, and update summarization tasks. While past work generally treats these problems differently ({\em e.g.}, different models are often used for generic and query-focused summarization), the submodular information measures allow us to study each of these problems via a unified approach. We first show that several previous query-focused and update summarization techniques have, unknowingly, used various instantiations of the aforesaid submodular information measures, providing evidence for the benefit and naturalness of these models. We then carefully study and demonstrate the modelling capabilities of the proposed functions in different settings and empirically verify our findings on both a synthetic dataset and an existing real-world image collection dataset (that has been extended by adding concept annotations to each image making it suitable for this task) and will be publicly released. We employ a max-margin framework to learn a mixture model built using the proposed instantiations of submodular information measures and demonstrate the effectiveness of our approach. While our experiments are in the context of image summarization, our framework is generic and can be easily extended to other summarization settings (e.g., videos or documents).

78.VMSMO: Learning to Generate Multimodal Summary for Video-based News Articles ⬇️

A popular multimedia news format nowadays is providing users with a lively video and a corresponding news article, which is employed by influential news media including CNN, BBC, and social media including Twitter and Weibo. In such a case, automatically choosing a proper cover frame of the video and generating an appropriate textual summary of the article can help editors save time, and readers make the decision more effectively. Hence, in this paper, we propose the task of Video-based Multimodal Summarization with Multimodal Output (VMSMO) to tackle such a problem. The main challenge in this task is to jointly model the temporal dependency of video with semantic meaning of article. To this end, we propose a Dual-Interaction-based Multimodal Summarizer (DIMS), consisting of a dual interaction module and multimodal generator. In the dual interaction module, we propose a conditional self-attention mechanism that captures local semantic information within video and a global-attention mechanism that handles the semantic relationship between news text and video from a high level. Extensive experiments conducted on a large-scale real-world VMSMO dataset show that DIMS achieves the state-of-the-art performance in terms of both automatic metrics and human evaluations.

79.RNN Training along Locally Optimal Trajectoriesvia Frank-Wolfe Algorithm ⬇️

We propose a novel and efficient training method for RNNs by iteratively seeking a local minima on the loss surface within a small region, and leverage this directional vector for the update, in an outer-loop. We propose to utilize the Frank-Wolfe (FW) algorithm in this context. Although, FW implicitly involves normalized gradients, which can lead to a slow convergence rate, we develop a novel RNN training method that, surprisingly, even with the additional cost, the overall training cost is empirically observed to be lower than back-propagation. Our method leads to a new Frank-Wolfe method, that is in essence an SGD algorithm with a restart scheme. We prove that under certain conditions our algorithm has a sublinear convergence rate of $O(1/\epsilon)$ for $\epsilon$ error. We then conduct empirical experiments on several benchmark datasets including those that exhibit long-term dependencies, and show significant performance improvement. We also experiment with deep RNN architectures and show efficient training performance. Finally, we demonstrate that our training method is robust to noisy data.

80.Miniscope3D: optimized single-shot miniature 3D fluorescence microscopy ⬇️

Miniature fluorescence microscopes are a standard tool in systems biology. However, widefield miniature microscopes capture only 2D information, and modifications that enable 3D capabilities increase the size and weight and have poor resolution outside a narrow depth range. Here, we achieve the 3D capability by replacing the tube lens of a conventional 2D Miniscope with an optimized multifocal phase mask at the objective's aperture stop. Placing the phase mask at the aperture stop significantly reduces the size of the device, and varying the focal lengths enables a uniform resolution across a wide depth range. The phase mask encodes the 3D fluorescence intensity into a single 2D measurement, and the 3D volume is recovered by solving a sparsity-constrained inverse problem. We provide methods for designing and fabricating the phase mask and an efficient forward model that accounts for the field-varying aberrations in miniature objectives. We demonstrate a prototype that is 17 mm tall and weighs 2.5 grams, achieving 2.76 $\mu$m lateral, and 15 $\mu$m axial resolution across most of the 900x700x390 $\mu m^3$ volume at 40 volumes per second. The performance is validated experimentally on resolution targets, dynamic biological samples, and mouse brain tissue. Compared with existing miniature single-shot volume-capture implementations, our system is smaller and lighter and achieves a more than 2x better lateral and axial resolution throughout a 10x larger usable depth range. Our microscope design provides single-shot 3D imaging for applications where a compact platform matters, such as volumetric neural imaging in freely moving animals and 3D motion studies of dynamic samples in incubators and lab-on-a-chip devices.

81.MAF: Multimodal Alignment Framework for Weakly-Supervised Phrase Grounding ⬇️

Phrase localization is a task that studies the mapping from textual phrases to regions of an image. Given difficulties in annotating phrase-to-object datasets at scale, we develop a Multimodal Alignment Framework (MAF) to leverage more widely-available caption-image datasets, which can then be used as a form of weak supervision. We first present algorithms to model phrase-object relevance by leveraging fine-grained visual representations and visually-aware language representations. By adopting a contrastive objective, our method uses information in caption-image pairs to boost the performance in weakly-supervised scenarios. Experiments conducted on the widely-adopted Flickr30k dataset show a significant improvement over existing weakly-supervised methods. With the help of the visually-aware language representations, we can also improve the previous best unsupervised result by 5.56%. We conduct ablation studies to show that both our novel model and our weakly-supervised strategies significantly contribute to our strong results.

82.A range characterization of the single-quadrant ADRT ⬇️

This work characterizes the range of the single-quadrant approximate discrete Radon transform (ADRT) of square images. The characterization is given in the form of linear constraints that ensure the exact and fast inversion formula [Rim, Appl. Math. Lett. 102 106159, 2020] yields a square image in a stable manner. The range characterization is obtained by first showing that the transform is a bijection between images supported on infinite half-strips, then identifying the linear subspaces that stay finitely supported under the inversion formula.

83.Domain Agnostic Learning for Unbiased Authentication ⬇️

Authentication is the task of confirming the matching relationship between a data instance and a given identity. Typical examples of authentication problems include face recognition and person re-identification. Data-driven authentication could be affected by undesired biases, i.e., the models are often trained in one domain (e.g., for people wearing spring outfits) while applied in other domains (e.g., they change the clothes to summer outfits). Previous works have made efforts to eliminate domain-difference. They typically assume domain annotations are provided, and all the domains share classes. However, for authentication, there could be a large number of domains shared by different identities/classes, and it is impossible to annotate these domains exhaustively. It could make domain-difference challenging to model and eliminate. In this paper, we propose a domain-agnostic method that eliminates domain-difference without domain labels. We alternately perform latent domain discovery and domain-difference elimination until our model no longer detects domain-difference. In our approach, the latent domains are discovered by learning the heterogeneous predictive relationships between inputs and outputs. Then domain-difference is eliminated in both class-dependent and class-independent components. Comprehensive empirical evaluation results are provided to demonstrate the effectiveness and superiority of our proposed method.

84.Towards Hardware-Agnostic Gaze-Trackers ⬇️

Gaze-tracking is a novel way of interacting with computers which allows new scenarios, such as enabling people with motor-neuron disabilities to control their computers or doctors to interact with patient information without touching screen or keyboard. Further, there are emerging applications of gaze-tracking in interactive gaming, user experience research, human attention analysis and behavioral studies. Accurate estimation of the gaze may involve accounting for head-pose, head-position, eye rotation, distance from the object as well as operating conditions such as illumination, occlusion, background noise and various biological aspects of the user. Commercially available gaze-trackers utilize specialized sensor assemblies that usually consist of an infrared light source and camera. There are several challenges in the universal proliferation of gaze-tracking as accessibility technologies, specifically its affordability, reliability, and ease-of-use. In this paper, we try to address these challenges through the development of a hardware-agnostic gaze-tracker. We present a deep neural network architecture as an appearance-based method for constrained gaze-tracking that utilizes facial imagery captured on an ordinary RGB camera ubiquitous in all modern computing devices. Our system achieved an error of 1.8073cm on GazeCapture dataset without any calibration or device specific fine-tuning. This research shows promise that one day soon any computer, tablet, or phone will be controllable using just your eyes due to the prediction capabilities of deep neutral networks.

85.Interpreting Multivariate Interactions in DNNs ⬇️

This paper aims to explain deep neural networks (DNNs) from the perspective of multivariate interactions. In this paper, we define and quantify the significance of interactions among multiple input variables of the DNN. Input variables with strong interactions usually form a coalition and reflect prototype features, which are memorized and used by the DNN for inference. We define the significance of interactions based on the Shapley value, which is designed to assign the attribution value of each input variable to the inference. We have conducted experiments with various DNNs. Experimental results have demonstrated the effectiveness of the proposed method.

86.Category-Learning with Context-Augmented Autoencoder ⬇️

Finding an interpretable non-redundant representation of real-world data is one of the key problems in Machine Learning. Biological neural networks are known to solve this problem quite well in unsupervised manner, yet unsupervised artificial neural networks either struggle to do it or require fine tuning for each task individually. We associate this with the fact that a biological brain learns in the context of the relationships between observations, while an artificial network does not. We also notice that, though a naive data augmentation technique can be very useful for supervised learning problems, autoencoders typically fail to generalize transformations from data augmentations. Thus, we believe that providing additional knowledge about relationships between data samples will improve model's capability of finding useful inner data representation. More formally, we consider a dataset not as a manifold, but as a category, where the examples are objects. Two these objects are connected by a morphism, if they actually represent different transformations of the same entity. Following this formalism, we propose a novel method of using data augmentations when training autoencoders. We train a Variational Autoencoder in such a way, that it makes transformation outcome predictable by auxiliary network in terms of the hidden representation. We believe that the classification accuracy of a linear classifier on the learned representation is a good metric to measure its interpretability. In our experiments, present approach outperforms $\beta$-VAE and is comparable with Gaussian-mixture VAE.

87.A Termination Criterion for Probabilistic PointClouds Registration ⬇️

Probabilistic Point Clouds Registration (PPCR) is an algorithm that, in its multi-iteration version, outperformed state of the art algorithms for local point clouds registration. However, its performances have been tested using a fixed high number of iterations. To be of practical usefulness, we think that the algorithm should decide by itself when to stop, to avoid an excessive number of iterations and, therefore, wasting computational time. With this work, we compare different termination criterion on several datasets and prove that the chosen one produce very good results that are comparable to those obtained using a very high number of iterations while saving computational time.

88.Double Forward Propagation for Memorized Batch Normalization ⬇️

Batch Normalization (BN) has been a standard component in designing deep neural networks (DNNs). Although the standard BN can significantly accelerate the training of DNNs and improve the generalization performance, it has several underlying limitations which may hamper the performance in both training and inference. In the training stage, BN relies on estimating the mean and variance of data using a single minibatch. Consequently, BN can be unstable when the batch size is very small or the data is poorly sampled. In the inference stage, BN often uses the so called moving mean and moving variance instead of batch statistics, i.e., the training and inference rules in BN are not consistent. Regarding these issues, we propose a memorized batch normalization (MBN), which considers multiple recent batches to obtain more accurate and robust statistics. Note that after the SGD update for each batch, the model parameters will change, and the features will change accordingly, leading to the Distribution Shift before and after the update for the considered batch. To alleviate this issue, we present a simple Double-Forward scheme in MBN which can further improve the performance. Compared to related methods, the proposed MBN exhibits consistent behaviors in both training and inference. Empirical results show that the MBN based models trained with the Double-Forward scheme greatly reduce the sensitivity of data and significantly improve the generalization performance.

89.An Empirical Study on Detecting COVID-19 in Chest X-ray Images Using Deep Learning Based Methods ⬇️

Spreading of COVID-19 virus has increased the efforts to provide testing kits. Not only the preparation of these kits had been hard, rare, and expensive but also using them is another issue. Results have shown that these kits take some crucial time to recognize the virus, in addition to the fact that they encounter with 30% loss. In this paper, we have studied the usage of x-ray pictures which are ubiquitous, for the classification of COVID-19 chest Xray images, by the existing convolutional neural networks (CNNs). We intend to train chest x-rays of infected and not infected ones with different CNNs architectures including VGG19, Densnet-121, and Xception. Training these architectures resulted in different accuracies which were much faster and more precise than usual ways of testing.

90.Contrastive Rendering for Ultrasound Image Segmentation ⬇️

Ultrasound (US) image segmentation embraced its significant improvement in deep learning era. However, the lack of sharp boundaries in US images still remains an inherent challenge for segmentation. Previous methods often resort to global context, multi-scale cues or auxiliary guidance to estimate the boundaries. It is hard for these methods to approach pixel-level learning for fine-grained boundary generating. In this paper, we propose a novel and effective framework to improve boundary estimation in US images. Our work has three highlights. First, we propose to formulate the boundary estimation as a rendering task, which can recognize ambiguous points (pixels/voxels) and calibrate the boundary prediction via enriched feature representation learning. Second, we introduce point-wise contrastive learning to enhance the similarity of points from the same class and contrastively decrease the similarity of points from different classes. Boundary ambiguities are therefore further addressed. Third, both rendering and contrastive learning tasks contribute to consistent improvement while reducing network parameters. As a proof-of-concept, we performed validation experiments on a challenging dataset of 86 ovarian US volumes. Results show that our proposed method outperforms state-of-the-art methods and has the potential to be used in clinical practice.

91.Unveiling Class-Labeling Structure for Universal Domain Adaptation ⬇️

As a more practical setting for unsupervised domain adaptation, Universal Domain Adaptation (UDA) is recently introduced, where the target label set is unknown. One of the big challenges in UDA is how to determine the common label set shared by source and target domains, as there is simply no labeling available in the target domain. In this paper, we employ a probabilistic approach for locating the common label set, where each source class may come from the common label set with a probability. In particular, we propose a novel approach for evaluating the probability of each source class from the common label set, where this probability is computed by the prediction margin accumulated over the whole target domain. Then, we propose a simple universal adaptation network (S-UAN) by incorporating the probabilistic structure for the common label set. Finally, we analyse the generalization bound focusing on the common label set and explore the properties on the target risk for UDA. Extensive experiments indicate that S-UAN works well in different UDA settings and outperforms the state-of-the-art methods by large margins.

92.Understanding Spatial Robustness of Deep Neural Networks ⬇️

Deep Neural Networks (DNNs) are being deployed in a wide range of settings today, from safety-critical applications like autonomous driving to commercial applications involving image classifications. However, recent research has shown that DNNs can be brittle to even slight variations of the input data. Therefore, rigorous testing of DNNs has gained widespread attention.
While DNN robustness under norm-bound perturbation got significant attention over the past few years, our knowledge is still limited when natural variants of the input images come. These natural variants, e.g. a rotated or a rainy version of the original input, are especially concerning as they can occur naturally in the field without any active adversary and may lead to undesirable consequences. Thus, it is important to identify the inputs whose small variations may lead to erroneous DNN behaviors. The very few studies that looked at DNN's robustness under natural variants, however, focus on estimating the overall robustness of DNNs across all the test data rather than localizing such error-producing points. This work aims to bridge this gap.
To this end, we study the local per-input robustness properties of the DNNs and leverage those properties to build a white-box (DEEPROBUST-W) and a black-box (DEEPROBUST-B) tool to automatically identify the non-robust points. Our evaluation of these methods on nine DNN models spanning three widely used image classification datasets shows that they are effective in flagging points of poor robustness. In particular, DEEPROBUST-W and DEEPROBUST-B are able to achieve an F1 score of up to 91.4% and 99.1%, respectively. We further show that DEEPROBUST-W can be applied to a regression problem for a self-driving car application.

93.Robust Behavioral Cloning for Autonomous Vehicles using End-to-End Imitation Learning ⬇️

In this work, we present a robust pipeline for cloning driving behavior of a human using end-to-end imitation learning. The proposed pipeline was employed to train and deploy three distinct driving behavior models onto a simulated vehicle. The training phase comprised of data collection, balancing, augmentation, preprocessing and training a neural network, following which, the trained model was deployed onto the ego vehicle to predict steering commands based on the feed from an onboard camera. A novel coupled control law was formulated to generate longitudinal control commands on-the-go based on the predicted steering angle and other parameters such as actual speed of the ego vehicle and the prescribed constraints for speed and steering. We analyzed computational efficiency of the pipeline and evaluated robustness of the trained models through exhaustive experimentation. Even a relatively shallow convolutional neural network model was able to learn key driving behaviors from sparsely labelled datasets and was tolerant to environmental variations during deployment of the said driving behaviors.