1.Anticipative Video Transformer ⬇️
We propose Anticipative Video Transformer (AVT), an end-to-end attention-based video modeling architecture that attends to the previously observed video in order to anticipate future actions. We train the model jointly to predict the next action in a video sequence, while also learning frame feature encoders that are predictive of successive future frames' features. Compared to existing temporal aggregation strategies, AVT has the advantage of both maintaining the sequential progression of observed actions while still capturing long-range dependencies--both critical for the anticipation task. Through extensive experiments, we show that AVT obtains the best reported performance on four popular action anticipation benchmarks: EpicKitchens-55, EpicKitchens-100, EGTEA Gaze+, and 50-Salads, including outperforming all submissions to the EpicKitchens-100 CVPR'21 challenge.
2.DynamicViT: Efficient Vision Transformers with Dynamic Token Sparsification ⬇️
Attention is sparse in vision transformers. We observe the final prediction in vision transformers is only based on a subset of most informative tokens, which is sufficient for accurate image recognition. Based on this observation, we propose a dynamic token sparsification framework to prune redundant tokens progressively and dynamically based on the input. Specifically, we devise a lightweight prediction module to estimate the importance score of each token given the current features. The module is added to different layers to prune redundant tokens hierarchically. To optimize the prediction module in an end-to-end manner, we propose an attention masking strategy to differentiably prune a token by blocking its interactions with other tokens. Benefiting from the nature of self-attention, the unstructured sparse tokens are still hardware friendly, which makes our framework easy to achieve actual speed-up. By hierarchically pruning 66% of the input tokens, our method greatly reduces 31%~37% FLOPs and improves the throughput by over 40% while the drop of accuracy is within 0.5% for various vision transformers. Equipped with the dynamic token sparsification framework, DynamicViT models can achieve very competitive complexity/accuracy trade-offs compared to state-of-the-art CNNs and vision transformers on ImageNet. Code is available at this https URL
3.Single Image Depth Estimation using Wavelet Decomposition ⬇️
We present a novel method for predicting accurate depths from monocular images with high efficiency. This optimal efficiency is achieved by exploiting wavelet decomposition, which is integrated in a fully differentiable encoder-decoder architecture. We demonstrate that we can reconstruct high-fidelity depth maps by predicting sparse wavelet coefficients. In contrast with previous works, we show that wavelet coefficients can be learned without direct supervision on coefficients. Instead we supervise only the final depth image that is reconstructed through the inverse wavelet transform. We additionally show that wavelet coefficients can be learned in fully self-supervised scenarios, without access to ground-truth depth. Finally, we apply our method to different state-of-the-art monocular depth estimation models, in each case giving similar or better results compared to the original model, while requiring less than half the multiply-adds in the decoder network. Code at this https URL
4.Neural Actor: Neural Free-view Synthesis of Human Actors with Pose Control ⬇️
We propose Neural Actor (NA), a new method for high-quality synthesis of humans from arbitrary viewpoints and under arbitrary controllable poses. Our method is built upon recent neural scene representation and rendering works which learn representations of geometry and appearance from only 2D images. While existing works demonstrated compelling rendering of static scenes and playback of dynamic scenes, photo-realistic reconstruction and rendering of humans with neural implicit methods, in particular under user-controlled novel poses, is still difficult. To address this problem, we utilize a coarse body model as the proxy to unwarp the surrounding 3D space into a canonical pose. A neural radiance field learns pose-dependent geometric deformations and pose- and view-dependent appearance effects in the canonical space from multi-view video input. To synthesize novel views of high fidelity dynamic geometry and appearance, we leverage 2D texture maps defined on the body model as latent variables for predicting residual deformations and the dynamic appearance. Experiments demonstrate that our method achieves better quality than the state-of-the-arts on playback as well as novel pose synthesis, and can even generalize well to new poses that starkly differ from the training poses. Furthermore, our method also supports body shape control of the synthesized results.
5.ProtoRes: Proto-Residual Architecture for Deep Modeling of Human Pose ⬇️
Our work focuses on the development of a learnable neural representation of human pose for advanced AI assisted animation tooling. Specifically, we tackle the problem of constructing a full static human pose based on sparse and variable user inputs (e.g. locations and/or orientations of a subset of body joints). To solve this problem, we propose a novel neural architecture that combines residual connections with prototype encoding of a partially specified pose to create a new complete pose from the learned latent space. We show that our architecture outperforms a baseline based on Transformer, both in terms of accuracy and computational efficiency. Additionally, we develop a user interface to integrate our neural model in Unity, a real-time 3D development platform. Furthermore, we introduce two new datasets representing the static human pose modeling problem, based on high-quality human motion capture data, which will be released publicly along with model code.
6.NeRFactor: Neural Factorization of Shape and Reflectance Under an Unknown Illumination ⬇️
We address the problem of recovering the shape and spatially-varying reflectance of an object from posed multi-view images of the object illuminated by one unknown lighting condition. This enables the rendering of novel views of the object under arbitrary environment lighting and editing of the object's material properties. The key to our approach, which we call Neural Radiance Factorization (NeRFactor), is to distill the volumetric geometry of a Neural Radiance Field (NeRF) [Mildenhall et al. 2020] representation of the object into a surface representation and then jointly refine the geometry while solving for the spatially-varying reflectance and the environment lighting. Specifically, NeRFactor recovers 3D neural fields of surface normals, light visibility, albedo, and Bidirectional Reflectance Distribution Functions (BRDFs) without any supervision, using only a re-rendering loss, simple smoothness priors, and a data-driven BRDF prior learned from real-world BRDF measurements. By explicitly modeling light visibility, NeRFactor is able to separate shadows from albedo and synthesize realistic soft or hard shadows under arbitrary lighting conditions. NeRFactor is able to recover convincing 3D models for free-viewpoint relighting in this challenging and underconstrained capture setup for both synthetic and real scenes. Qualitative and quantitative experiments show that NeRFactor outperforms classic and deep learning-based state of the art across various tasks. Our code and data are available at this http URL.
7.A Comparison for Anti-noise Robustness of Deep Learning Classification Methods on a Tiny Object Image Dataset: from Convolutional Neural Network to Visual Transformer and Performer ⬇️
Image classification has achieved unprecedented advance with the the rapid development of deep learning. However, the classification of tiny object images is still not well investigated. In this paper, we first briefly review the development of Convolutional Neural Network and Visual Transformer in deep learning, and introduce the sources and development of conventional noises and adversarial attacks. Then we use various models of Convolutional Neural Network and Visual Transformer to conduct a series of experiments on the image dataset of tiny objects (sperms and impurities), and compare various evaluation metrics in the experimental results to obtain a model with stable performance. Finally, we discuss the problems in the classification of tiny objects and make a prospect for the classification of tiny objects in the future.
8.You Never Cluster Alone ⬇️
Recent advances in self-supervised learning with instance-level contrastive objectives facilitate unsupervised clustering. However, a standalone datum is not perceiving the context of the holistic cluster, and may undergo sub-optimal assignment. In this paper, we extend the mainstream contrastive learning paradigm to a cluster-level scheme, where all the data subjected to the same cluster contribute to a unified representation that encodes the context of each data group. Contrastive learning with this representation then rewards the assignment of each datum. To implement this vision, we propose twin-contrast clustering (TCC). We define a set of categorical variables as clustering assignment confidence, which links the instance-level learning track with the cluster-level one. On one hand, with the corresponding assignment variables being the weight, a weighted aggregation along the data points implements the set representation of a cluster. We further propose heuristic cluster augmentation equivalents to enable cluster-level contrastive learning. On the other hand, we derive the evidence lower-bound of the instance-level contrastive objective with the assignments. By reparametrizing the assignment variables, TCC is trained end-to-end, requiring no alternating steps. Extensive experiments show that TCC outperforms the state-of-the-art on challenging benchmarks.
9.Adversarially Adaptive Normalization for Single Domain Generalization ⬇️
Single domain generalization aims to learn a model that performs well on many unseen domains with only one domain data for training. Existing works focus on studying the adversarial domain augmentation (ADA) to improve the model's generalization capability. The impact on domain generalization of the statistics of normalization layers is still underinvestigated. In this paper, we propose a generic normalization approach, adaptive standardization and rescaling normalization (ASR-Norm), to complement the missing part in previous works. ASR-Norm learns both the standardization and rescaling statistics via neural networks. This new form of normalization can be viewed as a generic form of the traditional normalizations. When trained with ADA, the statistics in ASR-Norm are learned to be adaptive to the data coming from different domains, and hence improves the model generalization performance across domains, especially on the target domain with large discrepancy from the source domain. The experimental results show that ASR-Norm can bring consistent improvement to the state-of-the-art ADA approaches by 1.6%, 2.7%, and 6.3% averagely on the Digits, CIFAR-10-C, and PACS benchmarks, respectively. As a generic tool, the improvement introduced by ASR-Norm is agnostic to the choice of ADA methods.
10.Learning High-Precision Bounding Box for Rotated Object Detection via Kullback-Leibler Divergence ⬇️
Existing rotated object detectors are mostly inherited from the horizontal detection paradigm, as the latter has evolved into a well-developed area. However, these detectors are difficult to perform prominently in high-precision detection due to the limitation of current regression loss design, especially for objects with large aspect ratios. Taking the perspective that horizontal detection is a special case for rotated object detection, in this paper, we are motivated to change the design of rotation regression loss from induction paradigm to deduction methodology, in terms of the relation between rotation and horizontal detection. We show that one essential challenge is how to modulate the coupled parameters in the rotation regression loss, as such the estimated parameters can influence to each other during the dynamic joint optimization, in an adaptive and synergetic way. Specifically, we first convert the rotated bounding box into a 2-D Gaussian distribution, and then calculate the Kullback-Leibler Divergence (KLD) between the Gaussian distributions as the regression loss. By analyzing the gradient of each parameter, we show that KLD (and its derivatives) can dynamically adjust the parameter gradients according to the characteristics of the object. It will adjust the importance (gradient weight) of the angle parameter according to the aspect ratio. This mechanism can be vital for high-precision detection as a slight angle error would cause a serious accuracy drop for large aspect ratios objects. More importantly, we have proved that KLD is scale invariant. We further show that the KLD loss can be degenerated into the popular
$l_{n}$ -norm loss for horizontal detection. Experimental results on seven datasets using different detectors show its consistent superiority, and codes are available at this https URL.
11.Robust Reference-based Super-Resolution via C2-Matching ⬇️
Reference-based Super-Resolution (Ref-SR) has recently emerged as a promising paradigm to enhance a low-resolution (LR) input image by introducing an additional high-resolution (HR) reference image. Existing Ref-SR methods mostly rely on implicit correspondence matching to borrow HR textures from reference images to compensate for the information loss in input images. However, performing local transfer is difficult because of two gaps between input and reference images: the transformation gap (e.g. scale and rotation) and the resolution gap (e.g. HR and LR). To tackle these challenges, we propose C2-Matching in this work, which produces explicit robust matching crossing transformation and resolution. 1) For the transformation gap, we propose a contrastive correspondence network, which learns transformation-robust correspondences using augmented views of the input image. 2) For the resolution gap, we adopt a teacher-student correlation distillation, which distills knowledge from the easier HR-HR matching to guide the more ambiguous LR-HR matching. 3) Finally, we design a dynamic aggregation module to address the potential misalignment issue. In addition, to faithfully evaluate the performance of Ref-SR under a realistic setting, we contribute the Webly-Referenced SR (WR-SR) dataset, mimicking the practical usage scenario. Extensive experiments demonstrate that our proposed C2-Matching significantly outperforms state of the arts by over 1dB on the standard CUFED5 benchmark. Notably, it also shows great generalizability on WR-SR dataset as well as robustness across large scale and rotation transformations.
12.Self-Supervised Learning of Event-Based Optical Flow with Spiking Neural Networks ⬇️
Neuromorphic sensing and computing hold a promise for highly energy-efficient and high-bandwidth-sensor processing. A major challenge for neuromorphic computing is that learning algorithms for traditional artificial neural networks (ANNs) do not transfer directly to spiking neural networks (SNNs) due to the discrete spikes and more complex neuronal dynamics. As a consequence, SNNs have not yet been successfully applied to complex, large-scale tasks. In this article, we focus on the self-supervised learning problem of optical flow estimation from event-based camera inputs, and investigate the changes that are necessary to the state-of-the-art ANN training pipeline in order to successfully tackle it with SNNs. More specifically, we first modify the input event representation to encode a much smaller time slice with minimal explicit temporal information. Consequently, we make the network's neuronal dynamics and recurrent connections responsible for integrating information over time. Moreover, we reformulate the self-supervised loss function for event-based optical flow to improve its convexity. We perform experiments with various types of recurrent ANNs and SNNs using the proposed pipeline. Concerning SNNs, we investigate the effects of elements such as parameter initialization and optimization, surrogate gradient shape, and adaptive neuronal mechanisms. We find that initialization and surrogate gradient width play a crucial part in enabling learning with sparse inputs, while the inclusion of adaptivity and learnable neuronal parameters can improve performance. We show that the performance of the proposed ANNs and SNNs are on par with that of the current state-of-the-art ANNs trained in a self-supervised manner.
13.E2E-VLP: End-to-End Vision-Language Pre-training Enhanced by Visual Learning ⬇️
Vision-language pre-training (VLP) on large-scale image-text pairs has achieved huge success for the cross-modal downstream tasks. The most existing pre-training methods mainly adopt a two-step training procedure, which firstly employs a pre-trained object detector to extract region-based visual features, then concatenates the image representation and text embedding as the input of Transformer to train. However, these methods face problems of using task-specific visual representation of the specific object detector for generic cross-modal understanding, and the computation inefficiency of two-stage pipeline. In this paper, we propose the first end-to-end vision-language pre-trained model for both V+L understanding and generation, namely E2E-VLP, where we build a unified Transformer framework to jointly learn visual representation, and semantic alignments between image and text. We incorporate the tasks of object detection and image captioning into pre-training with a unified Transformer encoder-decoder architecture for enhancing visual learning. An extensive set of experiments have been conducted on well-established vision-language downstream tasks to demonstrate the effectiveness of this novel VLP paradigm.
14.Less is More: Sparse Sampling for Dense Reaction Predictions ⬇️
Obtaining viewer responses from videos can be useful for creators and streaming platforms to analyze the video performance and improve the future user experience. In this report, we present our method for 2021 Evoked Expression from Videos Challenge. In particular, our model utilizes both audio and image modalities as inputs to predict emotion changes of viewers. To model long-range emotion changes, we use a GRU-based model to predict one sparse signal with 1Hz. We observe that the emotion changes are smooth. Therefore, the final dense prediction is obtained via linear interpolating the signal, which is robust to the prediction fluctuation. Albeit simple, the proposed method has achieved pearson's correlation score of 0.04430 on the final private test set.
15.Multi-Scale Feature Aggregation by Cross-Scale Pixel-to-Region Relation Operation for Semantic Segmentation ⬇️
Exploiting multi-scale features has shown great potential in tackling semantic segmentation problems. The aggregation is commonly done with sum or concatenation (concat) followed by convolutional (conv) layers. However, it fully passes down the high-level context to the following hierarchy without considering their interrelation. In this work, we aim to enable the low-level feature to aggregate the complementary context from adjacent high-level feature maps by a cross-scale pixel-to-region relation operation. We leverage cross-scale context propagation to make the long-range dependency capturable even by the high-resolution low-level features. To this end, we employ an efficient feature pyramid network to obtain multi-scale features. We propose a Relational Semantics Extractor (RSE) and Relational Semantics Propagator (RSP) for context extraction and propagation respectively. Then we stack several RSP into an RSP head to achieve the progressive top-down distribution of the context. Experiment results on two challenging datasets Cityscapes and COCO demonstrate that the RSP head performs competitively on both semantic segmentation and panoptic segmentation with high efficiency. It outperforms DeeplabV3 [1] by 0.7% with 75% fewer FLOPs (multiply-adds) in the semantic segmentation task.
16.GMAIR: Unsupervised Object Detection Based on Spatial Attention and Gaussian Mixture ⬇️
Recent studies on unsupervised object detection based on spatial attention have achieved promising results. Models, such as AIR and SPAIR, output "what" and "where" latent variables that represent the attributes and locations of objects in a scene, respectively. Most of the previous studies concentrate on the "where" localization performance; however, we claim that acquiring "what" object attributes is also essential for representation learning. This paper presents a framework, GMAIR, for unsupervised object detection. It incorporates spatial attention and a Gaussian mixture in a unified deep generative model. GMAIR can locate objects in a scene and simultaneously cluster them without supervision. Furthermore, we analyze the "what" latent variables and clustering process. Finally, we evaluate our model on MultiMNIST and Fruit2D datasets and show that GMAIR achieves competitive results on localization and clustering compared to state-of-the-art methods.
17.Towards urban scenes understanding through polarization cues ⬇️
Autonomous robotics is critically affected by the robustness of its scene understanding algorithms. We propose a two-axis pipeline based on polarization indices to analyze dynamic urban scenes. As robots evolve in unknown environments, they are prone to encountering specular obstacles. Usually, specular phenomena are rarely taken into account by algorithms which causes misinterpretations and erroneous estimates. By exploiting all the light properties, systems can greatly increase their robustness to events. In addition to the conventional photometric characteristics, we propose to include polarization sensing.
We demonstrate in this paper that the contribution of polarization measurement increases both the performances of segmentation and the quality of depth estimation. Our polarimetry-based approaches are compared here with other state-of-the-art RGB-centric methods showing interest of using polarization imaging.
18.Cross-Domain First Person Audio-Visual Action Recognition through Relative Norm Alignment ⬇️
First person action recognition is an increasingly researched topic because of the growing popularity of wearable cameras. This is bringing to light cross-domain issues that are yet to be addressed in this context. Indeed, the information extracted from learned representations suffers from an intrinsic environmental bias. This strongly affects the ability to generalize to unseen scenarios, limiting the application of current methods in real settings where trimmed labeled data are not available during training. In this work, we propose to leverage over the intrinsic complementary nature of audio-visual signals to learn a representation that works well on data seen during training, while being able to generalize across different domains. To this end, we introduce an audio-visual loss that aligns the contributions from the two modalities by acting on the magnitude of their feature norm representations. This new loss, plugged into a minimal multi-modal action recognition architecture, leads to strong results in cross-domain first person action recognition, as demonstrated by extensive experiments on the popular EPIC-Kitchens dataset.
19.APES: Audiovisual Person Search in Untrimmed Video ⬇️
Humans are arguably one of the most important subjects in video streams, many real-world applications such as video summarization or video editing workflows often require the automatic search and retrieval of a person of interest. Despite tremendous efforts in the person reidentification and retrieval domains, few works have developed audiovisual search strategies. In this paper, we present the Audiovisual Person Search dataset (APES), a new dataset composed of untrimmed videos whose audio (voices) and visual (faces) streams are densely annotated. APES contains over 1.9K identities labeled along 36 hours of video, making it the largest dataset available for untrimmed audiovisual person search. A key property of APES is that it includes dense temporal annotations that link faces to speech segments of the same identity. To showcase the potential of our new dataset, we propose an audiovisual baseline and benchmark for person retrieval. Our study shows that modeling audiovisual cues benefits the recognition of people's identities. To enable reproducibility and promote future research, the dataset annotations and baseline code are available at: this https URL
20.Generalized Domain Adaptation ⬇️
Many variants of unsupervised domain adaptation (UDA) problems have been proposed and solved individually. Its side effect is that a method that works for one variant is often ineffective for or not even applicable to another, which has prevented practical applications. In this paper, we give a general representation of UDA problems, named Generalized Domain Adaptation (GDA). GDA covers the major variants as special cases, which allows us to organize them in a comprehensive framework. Moreover, this generalization leads to a new challenging setting where existing methods fail, such as when domain labels are unknown, and class labels are only partially given to each domain. We propose a novel approach to the new setting. The key to our approach is self-supervised class-destructive learning, which enables the learning of class-invariant representations and domain-adversarial classifiers without using any domain labels. Extensive experiments using three benchmark datasets demonstrate that our method outperforms the state-of-the-art UDA methods in the new setting and that it is competitive in existing UDA variations as well.
21.Semantic Palette: Guiding Scene Generation with Class Proportions ⬇️
Despite the recent progress of generative adversarial networks (GANs) at synthesizing photo-realistic images, producing complex urban scenes remains a challenging problem. Previous works break down scene generation into two consecutive phases: unconditional semantic layout synthesis and image synthesis conditioned on layouts. In this work, we propose to condition layout generation as well for higher semantic control: given a vector of class proportions, we generate layouts with matching composition. To this end, we introduce a conditional framework with novel architecture designs and learning objectives, which effectively accommodates class proportions to guide the scene generation process. The proposed architecture also allows partial layout editing with interesting applications. Thanks to the semantic control, we can produce layouts close to the real distribution, helping enhance the whole scene generation process. On different metrics and urban scene benchmarks, our models outperform existing baselines. Moreover, we demonstrate the merit of our approach for data augmentation: semantic segmenters trained on real layout-image pairs along with additional ones generated by our approach outperform models only trained on real pairs.
22.Transferable Adversarial Examples for Anchor Free Object Detection ⬇️
Deep neural networks have been demonstrated to be vulnerable to adversarial attacks: subtle perturbation can completely change prediction result. The vulnerability has led to a surge of research in this direction, including adversarial attacks on object detection networks. However, previous studies are dedicated to attacking anchor-based object detectors. In this paper, we present the first adversarial attack on anchor-free object detectors. It conducts category-wise, instead of previously instance-wise, attacks on object detectors, and leverages high-level semantic information to efficiently generate transferable adversarial examples, which can also be transferred to attack other object detectors, even anchor-based detectors such as Faster R-CNN. Experimental results on two benchmark datasets demonstrate that our proposed method achieves state-of-the-art performance and transferability.
23.Imperceptible Adversarial Examples for Fake Image Detection ⬇️
Fooling people with highly realistic fake images generated with Deepfake or GANs brings a great social disturbance to our society. Many methods have been proposed to detect fake images, but they are vulnerable to adversarial perturbations -- intentionally designed noises that can lead to the wrong prediction. Existing methods of attacking fake image detectors usually generate adversarial perturbations to perturb almost the entire image. This is redundant and increases the perceptibility of perturbations. In this paper, we propose a novel method to disrupt the fake image detection by determining key pixels to a fake image detector and attacking only the key pixels, which results in the
$L_0$ and the$L_2$ norms of adversarial perturbations much less than those of existing works. Experiments on two public datasets with three fake image detectors indicate that our proposed method achieves state-of-the-art performance in both white-box and black-box attacks.
24.CT-Net: Channel Tensorization Network for Video Classification ⬇️
3D convolution is powerful for video classification but often computationally expensive, recent studies mainly focus on decomposing it on spatial-temporal and/or channel dimensions. Unfortunately, most approaches fail to achieve a preferable balance between convolutional efficiency and feature-interaction sufficiency. For this reason, we propose a concise and novel Channel Tensorization Network (CT-Net), by treating the channel dimension of input feature as a multiplication of K sub-dimensions. On one hand, it naturally factorizes convolution in a multiple dimension way, leading to a light computation burden. On the other hand, it can effectively enhance feature interaction from different channels, and progressively enlarge the 3D receptive field of such interaction to boost classification accuracy. Furthermore, we equip our CT-Module with a Tensor Excitation (TE) mechanism. It can learn to exploit spatial, temporal and channel attention in a high-dimensional manner, to improve the cooperative power of all the feature dimensions in our CT-Module. Finally, we flexibly adapt ResNet as our CT-Net. Extensive experiments are conducted on several challenging video benchmarks, e.g., Kinetics-400, Something-Something V1 and V2. Our CT-Net outperforms a number of recent SOTA approaches, in terms of accuracy and/or efficiency. The codes and models will be available on this https URL.
25.Attention-Guided Supervised Contrastive Learning for Semantic Segmentation ⬇️
Contrastive learning has shown superior performance in embedding global and spatial invariant features in computer vision (e.g., image classification). However, its overall success of embedding local and spatial variant features is still limited, especially for semantic segmentation. In a per-pixel prediction task, more than one label can exist in a single image for segmentation (e.g., an image contains both cat, dog, and grass), thereby it is difficult to define 'positive' or 'negative' pairs in a canonical contrastive learning setting. In this paper, we propose an attention-guided supervised contrastive learning approach to highlight a single semantic object every time as the target. With our design, the same image can be embedded to different semantic clusters with semantic attention (i.e., coerce semantic masks) as an additional input channel. To achieve such attention, a novel two-stage training strategy is presented. We evaluate the proposed method on multi-organ medical image segmentation task, as our major task, with both in-house data and BTCV 2015 datasets. Comparing with the supervised and semi-supervised training state-of-the-art in the backbone of ResNet-50, our proposed pipeline yields substantial improvement of 5.53% and 6.09% in Dice score for both medical image segmentation cohorts respectively. The performance of the proposed method on natural images is assessed via PASCAL VOC 2012 dataset, and achieves 2.75% substantial improvement.
26.Spline Positional Encoding for Learning 3D Implicit Signed Distance Fields ⬇️
Multilayer perceptrons (MLPs) have been successfully used to represent 3D shapes implicitly and compactly, by mapping 3D coordinates to the corresponding signed distance values or occupancy values. In this paper, we propose a novel positional encoding scheme, called Spline Positional Encoding, to map the input coordinates to a high dimensional space before passing them to MLPs, for helping to recover 3D signed distance fields with fine-scale geometric details from unorganized 3D point clouds. We verified the superiority of our approach over other positional encoding schemes on tasks of 3D shape reconstruction from input point clouds and shape space learning. The efficacy of our approach extended to image reconstruction is also demonstrated and evaluated.
27.When Vision Transformers Outperform ResNets without Pretraining or Strong Data Augmentations ⬇️
Vision Transformers (ViTs) and MLPs signal further efforts on replacing hand-wired features or inductive biases with general-purpose neural architectures. Existing works empower the models by massive data, such as large-scale pretraining and/or repeated strong data augmentations, and still report optimization-related problems (e.g., sensitivity to initialization and learning rate). Hence, this paper investigates ViTs and MLP-Mixers from the lens of loss geometry, intending to improve the models' data efficiency at training and generalization at inference. Visualization and Hessian reveal extremely sharp local minima of converged models. By promoting smoothness with a recently proposed sharpness-aware optimizer, we substantially improve the accuracy and robustness of ViTs and MLP-Mixers on various tasks spanning supervised, adversarial, contrastive, and transfer learning (e.g., +5.3% and +11.0% top-1 accuracy on ImageNet for ViT-B/16 and Mixer-B/16, respectively, with the simple Inception-style preprocessing). We show that the improved smoothness attributes to sparser active neurons in the first few layers. The resultant ViTs outperform ResNets of similar size and throughput when trained from scratch on ImageNet without large-scale pretraining or strong data augmentations. They also possess more perceptive attention maps.
28.SSMD: Semi-Supervised Medical Image Detection with Adaptive Consistency and Heterogeneous Perturbation ⬇️
Semi-Supervised classification and segmentation methods have been widely investigated in medical image analysis. Both approaches can improve the performance of fully-supervised methods with additional unlabeled data. However, as a fundamental task, semi-supervised object detection has not gained enough attention in the field of medical image analysis. In this paper, we propose a novel Semi-Supervised Medical image Detector (SSMD). The motivation behind SSMD is to provide free yet effective supervision for unlabeled data, by regularizing the predictions at each position to be consistent. To achieve the above idea, we develop a novel adaptive consistency cost function to regularize different components in the predictions. Moreover, we introduce heterogeneous perturbation strategies that work in both feature space and image space, so that the proposed detector is promising to produce powerful image representations and robust predictions. Extensive experimental results show that the proposed SSMD achieves the state-of-the-art performance at a wide range of settings. We also demonstrate the strength of each proposed module with comprehensive ablation studies.
29.Deconfounded Video Moment Retrieval with Causal Intervention ⬇️
We tackle the task of video moment retrieval (VMR), which aims to localize a specific moment in a video according to a textual query. Existing methods primarily model the matching relationship between query and moment by complex cross-modal interactions. Despite their effectiveness, current models mostly exploit dataset biases while ignoring the video content, thus leading to poor generalizability. We argue that the issue is caused by the hidden confounder in VMR, {i.e., temporal location of moments}, that spuriously correlates the model input and prediction. How to design robust matching models against the temporal location biases is crucial but, as far as we know, has not been studied yet for VMR.
To fill the research gap, we propose a causality-inspired VMR framework that builds structural causal model to capture the true effect of query and video content on the prediction. Specifically, we develop a Deconfounded Cross-modal Matching (DCM) method to remove the confounding effects of moment location. It first disentangles moment representation to infer the core feature of visual content, and then applies causal intervention on the disentangled multimodal input based on backdoor adjustment, which forces the model to fairly incorporate each possible location of the target into consideration. Extensive experiments clearly show that our approach can achieve significant improvement over the state-of-the-art methods in terms of both accuracy and generalization (Codes: \color{blue}{\url{this https URL}}
30.Noise Doesn't Lie: Towards Universal Detection of Deep Inpainting ⬇️
Deep image inpainting aims to restore damaged or missing regions in an image with realistic contents. While having a wide range of applications such as object removal and image recovery, deep inpainting techniques also have the risk of being manipulated for image forgery. A promising countermeasure against such forgeries is deep inpainting detection, which aims to locate the inpainted regions in an image. In this paper, we make the first attempt towards universal detection of deep inpainting, where the detection network can generalize well when detecting different deep inpainting methods. To this end, we first propose a novel data generation approach to generate a universal training dataset, which imitates the noise discrepancies exist in real versus inpainted image contents to train universal detectors. We then design a Noise-Image Cross-fusion Network (NIX-Net) to effectively exploit the discriminative information contained in both the images and their noise patterns. We empirically show, on multiple benchmark datasets, that our approach outperforms existing detection methods by a large margin and generalize well to unseen deep inpainting techniques. Our universal training dataset can also significantly boost the generalizability of existing detection methods.
31.Barbershop: GAN-based Image Compositing using Segmentation Masks ⬇️
Seamlessly blending features from multiple images is extremely challenging because of complex relationships in lighting, geometry, and partial occlusion which cause coupling between different parts of the image. Even though recent work on GANs enables synthesis of realistic hair or faces, it remains difficult to combine them into a single, coherent, and plausible image rather than a disjointed set of image patches. We present a novel solution to image blending, particularly for the problem of hairstyle transfer, based on GAN-inversion. We propose a novel latent space for image blending which is better at preserving detail and encoding spatial information, and propose a new GAN-embedding algorithm which is able to slightly modify images to conform to a common segmentation mask. Our novel representation enables the transfer of the visual properties from multiple reference images including specific details such as moles and wrinkles, and because we do image blending in a latent-space we are able to synthesize images that are coherent. Our approach avoids blending artifacts present in other approaches and finds a globally consistent image. Our results demonstrate a significant improvement over the current state of the art in a user study, with users preferring our blending solution over 95 percent of the time.
32.DeepCompress: Efficient Point Cloud Geometry Compression ⬇️
Point clouds are a basic data type that is increasingly of interest as 3D content becomes more ubiquitous. Applications using point clouds include virtual, augmented, and mixed reality and autonomous driving. We propose a more efficient deep learning-based encoder architecture for point clouds compression that incorporates principles from established 3D object detection and image compression architectures. Through an ablation study, we show that incorporating the learned activation function from Computational Efficient Neural Image Compression (CENIC) and designing more parameter-efficient convolutional blocks yields dramatic gains in efficiency and performance. Our proposed architecture incorporates Generalized Divisive Normalization activations and propose a spatially separable InceptionV4-inspired block. We then evaluate rate-distortion curves on the standard JPEG Pleno 8i Voxelized Full Bodies dataset to evaluate our model's performance. Our proposed modifications outperform the baseline approaches by a small margin in terms of Bjontegard delta rate and PSNR values, yet reduces necessary encoder convolution operations by 8 percent and reduces total encoder parameters by 20 percent. Our proposed architecture, when considered on its own, has a small penalty of 0.02 percent in Chamfer's Distance and 0.32 percent increased bit rate in Point to Plane Distance for the same peak signal-to-noise ratio.
33.Personalizing Pre-trained Models ⬇️
Self-supervised or weakly supervised models trained on large-scale datasets have shown sample-efficient transfer to diverse datasets in few-shot settings. We consider how upstream pretrained models can be leveraged for downstream few-shot, multilabel, and continual learning tasks. Our model CLIPPER (CLIP PERsonalized) uses image representations from CLIP, a large-scale image representation learning model trained using weak natural language supervision. We developed a technique, called Multi-label Weight Imprinting (MWI), for multi-label, continual, and few-shot learning, and CLIPPER uses MWI with image representations from CLIP. We evaluated CLIPPER on 10 single-label and 5 multi-label datasets. Our model shows robust and competitive performance, and we set new benchmarks for few-shot, multi-label, and continual learning. Our lightweight technique is also compute-efficient and enables privacy-preserving applications as the data is not sent to the upstream model for fine-tuning.
34.Multiscale Domain Adaptive YOLO for Cross-Domain Object Detection ⬇️
The area of domain adaptation has been instrumental in addressing the domain shift problem encountered by many applications. This problem arises due to the difference between the distributions of source data used for training in comparison with target data used during realistic testing scenarios. In this paper, we introduce a novel MultiScale Domain Adaptive YOLO (MS-DAYOLO) framework that employs multiple domain adaptation paths and corresponding domain classifiers at different scales of the recently introduced YOLOv4 object detector to generate domain-invariant features. We train and test our proposed method using popular datasets. Our experiments show significant improvements in object detection performance when training YOLOv4 using the proposed MS-DAYOLO and when tested on target data representing challenging weather conditions for autonomous driving applications.
35.Domain Adaptation for Facial Expression Classifier via Domain Discrimination and Gradient Reversal ⬇️
Bringing empathy to a computerized system could significantly improve the quality of human-computer communications, as soon as machines would be able to understand customer intentions and better serve their needs. According to different studies (Literature Review), visual information is one of the most important channels of human interaction and contains significant behavioral signals, that may be captured from facial expressions. Therefore, it is consistent and natural that the research in the field of Facial Expression Recognition (FER) has acquired increased interest over the past decade due to having diverse application area including health-care, sociology, psychology, driver-safety, virtual reality, cognitive sciences, security, entertainment, marketing, etc. We propose a new architecture for the task of FER and examine the impact of domain discrimination loss regularization on the learning process. With regard to observations, including both classical training conditions and unsupervised domain adaptation scenarios, important aspects of the considered domain adaptation approach integration are traced. The results may serve as a foundation for further research in the field.
36.NTIRE 2021 Challenge on High Dynamic Range Imaging: Dataset, Methods and Results ⬇️
This paper reviews the first challenge on high-dynamic range (HDR) imaging that was part of the New Trends in Image Restoration and Enhancement (NTIRE) workshop, held in conjunction with CVPR 2021. This manuscript focuses on the newly introduced dataset, the proposed methods and their results. The challenge aims at estimating a HDR image from one or multiple respective low-dynamic range (LDR) observations, which might suffer from under- or over-exposed regions and different sources of noise. The challenge is composed by two tracks: In Track 1 only a single LDR image is provided as input, whereas in Track 2 three differently-exposed LDR images with inter-frame motion are available. In both tracks, the ultimate goal is to achieve the best objective HDR reconstruction in terms of PSNR with respect to a ground-truth image, evaluated both directly and with a canonical tonemapping operation.
37.Unsharp Mask Guided Filtering ⬇️
The goal of this paper is guided image filtering, which emphasizes the importance of structure transfer during filtering by means of an additional guidance image. Where classical guided filters transfer structures using hand-designed functions, recent guided filters have been considerably advanced through parametric learning of deep networks. The state-of-the-art leverages deep networks to estimate the two core coefficients of the guided filter. In this work, we posit that simultaneously estimating both coefficients is suboptimal, resulting in halo artifacts and structure inconsistencies. Inspired by unsharp masking, a classical technique for edge enhancement that requires only a single coefficient, we propose a new and simplified formulation of the guided filter. Our formulation enjoys a filtering prior from a low-pass filter and enables explicit structure transfer by estimating a single coefficient. Based on our proposed formulation, we introduce a successive guided filtering network, which provides multiple filtering results from a single network, allowing for a trade-off between accuracy and efficiency. Extensive ablations, comparisons and analysis show the effectiveness and efficiency of our formulation and network, resulting in state-of-the-art results across filtering tasks like upsampling, denoising, and cross-modality filtering. Code is available at \url{this https URL}.
38.Learning to Select: A Fully Attentive Approach for Novel Object Captioning ⬇️
Image captioning models have lately shown impressive results when applied to standard datasets. Switching to real-life scenarios, however, constitutes a challenge due to the larger variety of visual concepts which are not covered in existing training sets. For this reason, novel object captioning (NOC) has recently emerged as a paradigm to test captioning models on objects which are unseen during the training phase. In this paper, we present a novel approach for NOC that learns to select the most relevant objects of an image, regardless of their adherence to the training set, and to constrain the generative process of a language model accordingly. Our architecture is fully-attentive and end-to-end trainable, also when incorporating constraints. We perform experiments on the held-out COCO dataset, where we demonstrate improvements over the state of the art, both in terms of adaptability to novel objects and caption quality.
39.Container: Context Aggregation Network ⬇️
Convolutional neural networks (CNNs) are ubiquitous in computer vision, with a myriad of effective and efficient variations. Recently, Transformers -- originally introduced in natural language processing -- have been increasingly adopted in computer vision. While early adopters continue to employ CNN backbones, the latest networks are end-to-end CNN-free Transformer solutions. A recent surprising finding shows that a simple MLP based solution without any traditional convolutional or Transformer components can produce effective visual representations. While CNNs, Transformers and MLP-Mixers may be considered as completely disparate architectures, we provide a unified view showing that they are in fact special cases of a more general method to aggregate spatial context in a neural network stack. We present the \model (CONText AggregatIon NEtwoRk), a general-purpose building block for multi-head context aggregation that can exploit long-range interactions \emph{a la} Transformers while still exploiting the inductive bias of the local convolution operation leading to faster convergence speeds, often seen in CNNs. In contrast to Transformer-based methods that do not scale well to downstream tasks that rely on larger input image resolutions, our efficient network, named \modellight, can be employed in object detection and instance segmentation networks such as DETR, RetinaNet and Mask-RCNN to obtain an impressive detection mAP of 38.9, 43.8, 45.1 and mask mAP of 41.3, providing large improvements of 6.6, 7.3, 6.9 and 6.6 pts respectively, compared to a ResNet-50 backbone with a comparable compute and parameter size. Our method also achieves promising results on self-supervised learning compared to DeiT on the DINO framework.
40.Convolutional Neural Network(CNN/ConvNet) in Stock Price Movement Prediction ⬇️
With technological advancements and the exponential growth of data, we have been unfolding different capabilities of neural networks in different sectors. In this paper, I have tried to use a specific type of Neural Network known as Convolutional Neural Network(CNN/ConvNet) in the stock market. In other words, I have tried to construct and train a convolutional neural network on past stock prices data and then tried to predict the movement of stock price i.e. whether the stock price would rise or fall, in the coming time.
41.Pathology-Aware Generative Adversarial Networks for Medical Image Augmentation ⬇️
Convolutional Neural Networks (CNNs) can play a key role in Medical Image Analysis under large-scale annotated datasets. However, preparing such massive dataset is demanding. In this context, Generative Adversarial Networks (GANs) can generate realistic but novel samples, and thus effectively cover the real image distribution. In terms of interpolation, the GAN-based medical image augmentation is reliable because medical modalities can display the human body's strong anatomical consistency at fixed position while clearly reflecting inter-subject variability; thus, we propose to use noise-to-image GANs (e.g., random noise samples to diverse pathological images) for (i) medical Data Augmentation (DA) and (ii) physician training. Regarding the DA, the GAN-generated images can improve Computer-Aided Diagnosis based on supervised learning. For the physician training, the GANs can display novel desired pathological images and help train medical trainees despite infrastructural/legal constraints. This thesis contains four GAN projects aiming to present such novel applications' clinical relevance in collaboration with physicians. Whereas the methods are more generally applicable, this thesis only explores a few oncological applications.
42.Robotic Inspection and 3D GPR-based Reconstruction for Underground Utilities ⬇️
Ground Penetrating Radar (GPR) is an effective non-destructive evaluation (NDE) device for inspecting and surveying subsurface objects (i.e., rebars, utility pipes) in complex environments. However, the current practice for GPR data collection requires a human inspector to move a GPR cart along pre-marked grid lines and record the GPR data in both X and Y directions for post-processing by 3D GPR imaging software. It is time-consuming and tedious work to survey a large area. Furthermore, identifying the subsurface targets depends on the knowledge of an experienced engineer, who has to make manual and subjective interpretation that limits the GPR applications, especially in large-scale scenarios. In addition, the current GPR imaging technology is not intuitive, and not for normal users to understand, and not friendly to visualize. To address the above challenges, this paper presents a novel robotic system to collect GPR data, interpret GPR data, localize the underground utilities, reconstruct and visualize the underground objects' dense point cloud model in a user-friendly manner. This system is composed of three modules: 1) a vision-aided Omni-directional robotic data collection platform, which enables the GPR antenna to scan the target area freely with an arbitrary trajectory while using a visual-inertial-based positioning module tags the GPR measurements with positioning information; 2) a deep neural network (DNN) migration module to interpret the raw GPR B-scan image into a cross-section of object model; 3) a DNN-based 3D reconstruction method, i.e., GPRNet, to generate underground utility model represented as fine 3D point cloud. Comparative studies on synthetic and field GPR raw data with various incompleteness and noise are performed.
43.Denoising and Optical and SAR Image Classifications Based on Feature Extraction and Sparse Representation ⬇️
Optical image data have been used by the Remote Sensing workforce to study land use and cover since such data is easily interpretable. Synthetic Aperture Radar (SAR) has the characteristic of obtaining images during all-day, all-weather and provides object information that is different from visible and infrared sensors. However, SAR images have more speckle noise and fewer dimensions. This paper presents a method for denoising, feature extraction and compares classifications of Optical and SAR images. The image was denoised using K-Singular Value Decomposition (K-SVD) algorithm. A method to map the extraordinary goal signatures to be had withinside the SAR or Optical image using support vector machine (SVM) through offering given the enter facts to the supervised classifier. Initially, the Gray Level Histogram (GLH) and Gray Level Co-occurrence Matrix (GLCM) are used for feature extraction. Secondly, the extracted feature vectors from the first step were combined using correlation analysis to reduce the dimensionality of the feature spaces. Thirdly, the Classification of SAR images was done in Sparse Representations Classification (SRC). The above-mentioned classifications techniques were developed and performance parameters are accuracy and Kappa Coefficient calculated using MATLAB 2018a.
44.Simultaneous Multi-View Object Recognition and Grasping in Open-Ended Domains ⬇️
A robot working in human-centric environments needs to know which kind of objects exist in the scene, where they are, and how to grasp and manipulate various objects in different situations to help humans in everyday tasks. Therefore, object recognition and grasping are two key functionalities for such robots. Most state-of-the-art tackles object recognition and grasping as two separate problems while both use visual input. Furthermore, the knowledge of the robot is fixed after the training phase. In such cases, if the robot faces new object categories, it must retrain from scratch to incorporate new information without catastrophic interference. To address this problem, we propose a deep learning architecture with augmented memory capacities to handle open-ended object recognition and grasping simultaneously. In particular, our approach takes multi-views of an object as input and jointly estimates pixel-wise grasp configuration as well as a deep scale- and rotation-invariant representation as outputs. The obtained representation is then used for open-ended object recognition through a meta-active learning technique. We demonstrate the ability of our approach to grasp never-seen-before objects and to rapidly learn new object categories using very few examples on-site in both simulation and real-world settings.
45.Separated-Spectral-Distribution Estimation Based on Bayesian Inference with Single RGB Camera ⬇️
In this paper, we propose a novel method for separately estimating spectral distributions from images captured by a typical RGB camera. The proposed method allows us to separately estimate a spectral distribution of illumination, reflectance, or camera sensitivity, while recent hyperspectral cameras are limited to capturing a joint spectral distribution from a scene. In addition, the use of Bayesian inference makes it possible to take into account prior information of both spectral distributions and image noise as probability distributions. As a result, the proposed method can estimate spectral distributions in a unified way, and it can enhance the robustness of the estimation against noise, which conventional spectral-distribution estimation methods cannot. The use of Bayesian inference also enables us to obtain the confidence of estimation results. In an experiment, the proposed method is shown not only to outperform conventional estimation methods in terms of RMSE but also to be robust against noise.
46.Noisy Labels are Treasure: Mean-Teacher-Assisted Confident Learning for Hepatic Vessel Segmentation ⬇️
Manually segmenting the hepatic vessels from Computer Tomography (CT) is far more expertise-demanding and laborious than other structures due to the low-contrast and complex morphology of vessels, resulting in the extreme lack of high-quality labeled data. Without sufficient high-quality annotations, the usual data-driven learning-based approaches struggle with deficient training. On the other hand, directly introducing additional data with low-quality annotations may confuse the network, leading to undesirable performance degradation. To address this issue, we propose a novel mean-teacher-assisted confident learning framework to robustly exploit the noisy labeled data for the challenging hepatic vessel segmentation task. Specifically, with the adapted confident learning assisted by a third party, i.e., the weight-averaged teacher model, the noisy labels in the additional low-quality dataset can be transformed from "encumbrance" to "treasure" via progressive pixel-wise soft-correction, thus providing productive guidance. Extensive experiments using two public datasets demonstrate the superiority of the proposed framework as well as the effectiveness of each component.
47.Deep Learning Based Analysis of Prostate Cancer from MP-MRI ⬇️
The diagnosis of prostate cancer faces a problem with overdiagnosis that leads to damaging side effects due to unnecessary treatment. Research has shown that the use of multi-parametric magnetic resonance images to conduct biopsies can drastically help to mitigate the overdiagnosis, thus reducing the side effects on healthy patients. This study aims to investigate the use of deep learning techniques to explore computer-aid diagnosis based on MRI as input. Several diagnosis problems ranging from classification of lesions as being clinically significant or not to the detection and segmentation of lesions are addressed with deep learning based approaches.
This thesis tackled two main problems regarding the diagnosis of prostate cancer. Firstly, XmasNet was used to conduct two large experiments on the classification of lesions. Secondly, detection and segmentation experiments were conducted, first on the prostate and afterward on the prostate cancer lesions. The former experiments explored the lesions through a two-dimensional space, while the latter explored models to work with three-dimensional inputs. For this task, the 3D models explored were the 3D U-Net and a pretrained 3D ResNet-18. A rigorous analysis of all these problems was conducted with a total of two networks, two cropping techniques, two resampling techniques, two crop sizes, five input sizes and data augmentations experimented for lesion classification. While for segmentation two models, two input sizes and data augmentations were experimented. However, while the binary classification of the clinical significance of lesions and the detection and segmentation of the prostate already achieve the desired results (0.870 AUC and 0.915 dice score respectively), the classification of the PIRADS score and the segmentation of lesions still have a large margin to improve (0.664 accuracy and 0.690 dice score respectively).
48.Effort-free Automated Skeletal Abnormality Detection of Rat Fetuses on Whole-body Micro-CT Scans ⬇️
Machine Learning-based fast and quantitative automated screening plays a key role in analyzing human bones on Computed Tomography (CT) scans. However, despite the requirement in drug safety assessment, such research is rare on animal fetus micro-CT scans due to its laborious data collection and annotation. Therefore, we propose various bone feature engineering techniques to thoroughly automate the skeletal localization/labeling/abnormality detection of rat fetuses on whole-body micro-CT scans with minimum effort. Despite limited training data of 49 fetuses, in skeletal labeling and abnormality detection, we achieve accuracy of 0.900 and 0.810, respectively.
49.Partial Graph Reasoning for Neural Network Regularization ⬇️
Regularizers helped deep neural networks prevent feature co-adaptations. Dropout,as a commonly used regularization technique, stochastically disables neuron ac-tivations during network optimization. However, such complete feature disposal can affect the feature representation and network understanding. Toward betterdescriptions of latent representations, we present DropGraph that learns regularization function by constructing a stand-alone graph from the backbone features. DropGraph first samples stochastic spatial feature vectors and then incorporates graph reasoning methods to generate feature map distortions. This add-on graph regularizes the network during training and can be completely skipped during inference. We provide intuitions on the linkage between graph reasoning andDropout with further discussions on how partial graph reasoning method reduces feature correlations. To this end, we extensively study the modeling of graphvertex dependencies and the utilization of the graph for distorting backbone featuremaps. DropGraph was validated on four tasks with a total of 7 different datasets.The experimental results show that our method outperforms other state-of-the-art regularizers while leaving the base model structure unmodified during inference.
50.Advances in Classifying the Stages of Diabetic Retinopathy Using Convolutional Neural Networks in Low Memory Edge Devices ⬇️
Diabetic Retinopathy (DR) is a severe complication that may lead to retinal vascular damage and is one of the leading causes of vision impairment and blindness. DR broadly is classified into two stages - non-proliferative (NPDR), where there are almost no symptoms, except a few microaneurysms, and proliferative (PDR) involving a huge number of microaneurysms and hemorrhages, soft and hard exudates, neo-vascularization, macular ischemia or a combination of these, making it easier to detect. More specifically, DR is usually classified into five levels, labeled 0-4, from 0 indicating no DR to 4 which is most severe. This paper firstly presents a discussion on the risk factors of the disease, then surveys the recent literature on the topic followed by examining certain techniques which were found to be highly effective in improving the prognosis accuracy. Finally, a convolutional neural network model is proposed to detect all the stages of DR on a low-memory edge microcontroller. The model has a size of just 5.9 MB, accuracy and F1 score both of 94% and an inference speed of about 20 frames per second.
51.Fast improvement of TEM image with low-dose electrons by deep learning ⬇️
Low-electron-dose observation is indispensable for observing various samples using a transmission electron microscope; consequently, image processing has been used to improve transmission electron microscopy (TEM) images. To apply such image processing to in situ observations, we here apply a convolutional neural network to TEM imaging. Using a dataset that includes short-exposure images and long-exposure images, we develop a pipeline for processed short-exposure images, based on end-to-end training. The quality of images acquired with a total dose of approximately 5 e- per pixel becomes comparable to that of images acquired with a total dose of approximately 1000 e- per pixel. Because the conversion time is approximately 8 ms, in situ observation at 125 fps is possible. This imaging technique enables in situ observation of electron-beam-sensitive specimens.
52.Machine Learning Based Texture Analysis of Patella from X-Rays for Detecting Patellofemoral Osteoarthritis ⬇️
Objective is to assess the ability of texture features for detecting radiographic patellofemoral osteoarthritis (PFOA) from knee lateral view radiographs. We used lateral view knee radiographs from MOST public use datasets (n = 5507 knees). Patellar region-of-interest (ROI) was automatically detected using landmark detection tool (BoneFinder). Hand-crafted features, based on LocalBinary Patterns (LBP), were then extracted to describe the patellar texture. First, a machine learning model (Gradient Boosting Machine) was trained to detect radiographic PFOA from the LBP features. Furthermore, we used end-to-end trained deep convolutional neural networks (CNNs) directly on the texture patches for detecting the PFOA. The proposed classification models were eventually compared with more conventional reference models that use clinical assessments and participant characteristics such as age, sex, body mass index(BMI), the total WOMAC score, and tibiofemoral Kellgren-Lawrence (KL) grade. Atlas-guided visual assessment of PFOA status by expert readers provided in the MOST public use datasets was used as a classification outcome for the models. Performance of prediction models was assessed using the area under the receiver operating characteristic curve (ROC AUC), the area under the precision-recall (PR) curve-average precision (AP)-, and Brier score in the stratified 5-fold cross validation setting.Of the 5507 knees, 953 (17.3%) had PFOA. AUC and AP for the strongest reference model including age, sex, BMI, WOMAC score, and tibiofemoral KL grade to predict PFOA were 0.817 and 0.487, respectively. Textural ROI classification using CNN significantly improved the prediction performance (ROC AUC= 0.889, AP= 0.714). We present the first study that analyses patellar bone texture for diagnosing PFOA. Our results demonstrates the potential of using texture features of patella to predict PFOA.
53.Improving the Transferability of Adversarial Examples with New Iteration Framework and Input Dropout ⬇️
Deep neural networks(DNNs) is vulnerable to be attacked by adversarial examples. Black-box attack is the most threatening attack. At present, black-box attack methods mainly adopt gradient-based iterative attack methods, which usually limit the relationship between the iteration step size, the number of iterations, and the maximum perturbation. In this paper, we propose a new gradient iteration framework, which redefines the relationship between the above three. Under this framework, we easily improve the attack success rate of DI-TI-MIM. In addition, we propose a gradient iterative attack method based on input dropout, which can be well combined with our framework. We further propose a multi dropout rate version of this method. Experimental results show that our best method can achieve attack success rate of 96.2% for defense model on average, which is higher than the state-of-the-art gradient-based attacks.
54.Grounding Complex Navigational Instructions Using Scene Graphs ⬇️
Training a reinforcement learning agent to carry out natural language instructions is limited by the available supervision, i.e. knowing when the instruction has been carried out. We adapt the CLEVR visual question answering dataset to generate complex natural language navigation instructions and accompanying scene graphs, yielding an environment-agnostic supervised dataset. To demonstrate the use of this data set, we map the scenes to the VizDoom environment and use the architecture in \citet{gatedattention} to train an agent to carry out these more complex language instructions.
55.Exploring Memorization in Adversarial Training ⬇️
It is well known that deep learning models have a propensity for fitting the entire training set even with random labels, which requires memorization of every training sample. In this paper, we investigate the memorization effect in adversarial training (AT) for promoting a deeper understanding of capacity, convergence, generalization, and especially robust overfitting of adversarially trained classifiers. We first demonstrate that deep networks have sufficient capacity to memorize adversarial examples of training data with completely random labels, but not all AT algorithms can converge under the extreme circumstance. Our study of AT with random labels motivates further analyses on the convergence and generalization of AT. We find that some AT methods suffer from a gradient instability issue, and the recently suggested complexity measures cannot explain robust generalization by considering models trained on random labels. Furthermore, we identify a significant drawback of memorization in AT that it could result in robust overfitting. We then propose a new mitigation algorithm motivated by detailed memorization analyses. Extensive experiments on various datasets validate the effectiveness of the proposed method.
56.PDPGD: Primal-Dual Proximal Gradient Descent Adversarial Attack ⬇️
State-of-the-art deep neural networks are sensitive to small input perturbations. Since the discovery of this intriguing vulnerability, many defence methods have been proposed that attempt to improve robustness to adversarial noise. Fast and accurate attacks are required to compare various defence methods. However, evaluating adversarial robustness has proven to be extremely challenging. Existing norm minimisation adversarial attacks require thousands of iterations (e.g. Carlini & Wagner attack), are limited to the specific norms (e.g. Fast Adaptive Boundary), or produce sub-optimal results (e.g. Brendel & Bethge attack). On the other hand, PGD attack, which is fast, general and accurate, ignores the norm minimisation penalty and solves a simpler perturbation-constrained problem. In this work, we introduce a fast, general and accurate adversarial attack that optimises the original non-convex constrained minimisation problem. We interpret optimising the Lagrangian of the adversarial attack optimisation problem as a two-player game: the first player minimises the Lagrangian wrt the adversarial noise; the second player maximises the Lagrangian wrt the regularisation penalty. Our attack algorithm simultaneously optimises primal and dual variables to find the minimal adversarial perturbation. In addition, for non-smooth
$l_p$ -norm minimisation, such as$l_{\infty}$ -,$l_1$ -, and$l_0$ -norms, we introduce primal-dual proximal gradient descent attack. We show in the experiments that our attack outperforms current state-of-the-art$l_{\infty}$ -,$l_2$ -,$l_1$ -, and$l_0$ -attacks on MNIST, CIFAR-10 and Restricted ImageNet datasets against unregularised and adversarially trained models.
57.Not All Knowledge Is Created Equal ⬇️
Mutual knowledge distillation (MKD) improves a model by distilling knowledge from another model. However, not all knowledge is certain and correct, especially under adverse conditions. For example, label noise usually leads to less reliable models due to the undesired memorisation [1, 2]. Wrong knowledge misleads the learning rather than helps. This problem can be handled by two aspects: (i) improving the reliability of a model where the knowledge is from (i.e., knowledge source's reliability); (ii) selecting reliable knowledge for distillation. In the literature, making a model more reliable is widely studied while selective MKD receives little attention. Therefore, we focus on studying selective MKD and highlight its importance in this work.
Concretely, a generic MKD framework, Confident knowledge selection followed by Mutual Distillation (CMD), is designed. The key component of CMD is a generic knowledge selection formulation, making the selection threshold either static (CMD-S) or progressive (CMD-P). Additionally, CMD covers two special cases: zero knowledge and all knowledge, leading to a unified MKD framework. We empirically find CMD-P performs better than CMD-S. The main reason is that a model's knowledge upgrades and becomes confident as the training progresses.
Extensive experiments are present to demonstrate the effectiveness of CMD and thoroughly justify the design of CMD. For example, CMD-P obtains new state-of-the-art results in robustness against label noise.
58.LLC: Accurate, Multi-purpose Learnt Low-dimensional Binary Codes ⬇️
Learning binary representations of instances and classes is a classical problem with several high potential applications. In modern settings, the compression of high-dimensional neural representations to low-dimensional binary codes is a challenging task and often require large bit-codes to be accurate. In this work, we propose a novel method for Learning Low-dimensional binary Codes (LLC) for instances as well as classes. Our method does not require any side-information, like annotated attributes or label meta-data, and learns extremely low-dimensional binary codes (~20 bits for ImageNet-1K). The learnt codes are super-efficient while still ensuring nearly optimal classification accuracy for ResNet50 on ImageNet-1K. We demonstrate that the learnt codes capture intrinsically important features in the data, by discovering an intuitive taxonomy over classes. We further quantitatively measure the quality of our codes by applying it to the efficient image retrieval as well as out-of-distribution (OOD) detection problems. For ImageNet-100 retrieval problem, our learnt binary codes outperform 16 bit HashNet using only 10 bits and also are as accurate as 10 dimensional real representations. Finally, our learnt binary codes can perform OOD detection, out-of-the-box, as accurately as a baseline that needs ~3000 samples to tune its threshold, while we require none. Code and pre-trained models are available at this https URL.
59.SMURF: SeMantic and linguistic UndeRstanding Fusion for Caption Evaluation via Typicality Analysis ⬇️
The open-ended nature of visual captioning makes it a challenging area for evaluation. The majority of proposed models rely on specialized training to improve human-correlation, resulting in limited adoption, generalizability, and explainabilty. We introduce "typicality", a new formulation of evaluation rooted in information theory, which is uniquely suited for problems lacking a definite ground truth. Typicality serves as our framework to develop a novel semantic comparison, SPARCS, as well as referenceless fluency evaluation metrics. Over the course of our analysis, two separate dimensions of fluency naturally emerge: style, captured by metric SPURTS, and grammar, captured in the form of grammatical outlier penalties. Through extensive experiments and ablation studies on benchmark datasets, we show how these decomposed dimensions of semantics and fluency provide greater system-level insight into captioner differences. Our proposed metrics along with their combination, SMURF, achieve state-of-the-art correlation with human judgment when compared with other rule-based evaluation metrics.
60.One Representation to Rule Them All: Identifying Out-of-Support Examples in Few-shot Learning with Generic Representations ⬇️
The field of few-shot learning has made remarkable strides in developing powerful models that can operate in the small data regime. Nearly all of these methods assume every unlabeled instance encountered will belong to a handful of known classes for which one has examples. This can be problematic for real-world use cases where one routinely finds 'none-of-the-above' examples. In this paper we describe this challenge of identifying what we term 'out-of-support' (OOS) examples. We describe how this problem is subtly different from out-of-distribution detection and describe a new method of identifying OOS examples within the Prototypical Networks framework using a fixed point which we call the generic representation. We show that our method outperforms other existing approaches in the literature as well as other approaches that we propose in this paper. Finally, we investigate how the use of such a generic point affects the geometry of a model's feature space.