CVPR 2020 论文列表

Advanced object detectors usually adopt a backbone network designed and pretrained by ImageNet classification. Recently neural architecture search (NAS) has emerged to automatically design a task-specific backbone to bridge the gap between the tasks of classification and detection. In this paper, we propose a two-phase serial-to-parallel architecture search framework named SP-NAS towards a flexible task-oriented detection backbone. Specifically, the serial-searching round aims at finding a sequence of serial blocks with optimal scale and output channels in the feature hierarchy by a Swap-Expand-Reignite search algorithm; the parallel-searching phase then assembles several sub-architectures along with the previous searched backbone into a more powerful parallel-structured backbone. We efficiently search a detection backbone by exploring a network morphism strategy on multiple detection benchmarks. The resulting architectures achieve SOTA results, i.e. top performance (LAMR: 0.055) on the automotive detection leaderboard of EuroCityPersons benchmark, improving 2.3% mAP with less FLOPS than NAS-FPN on COCO, and reaching 84.1% AP50 on VOC better than DetNAS and Auto-FPN in terms of both accuracy and speed.

Based on the framework of multiple instance learning (MIL), tremendous works have promoted the advances of weakly supervised object detection (WSOD). However, most MIL-based methods tend to localize instances to their discriminative parts instead of the whole content. In this paper, we propose a spatial likelihood voting (SLV) module to converge the proposal localizing process without any bounding box annotations. Specifically, all region proposals in a given image play the role of voters every iteration during training, voting for the likelihood of each category in spatial dimensions. After dilating alignment on the area with large likelihood values, the voting results are regularized as bounding boxes, being used for the final classification and localization. Based on SLV, we further propose an end-to-end training framework for multi-task learning. The classification and localization tasks promote each other, which further improves the detection performance. Extensive experiments on the PASCAL VOC 2007 and 2012 datasets demonstrate the superior performance of SLV.

Architecture design has become a crucial component of successful deep learning. Recent progress in automatic neural architecture search (NAS) shows a lot of promise. However, discovered architectures often fail to generalize in the final evaluation. Architectures with a higher validation accuracy during the search phase may perform worse in the evaluation. Aiming to alleviate this common issue, we introduce sequential greedy architecture search (SGAS), an efficient method for neural architecture search. By dividing the search procedure into sub-problems, SGAS chooses and prunes candidate operations in a greedy fashion. We apply SGAS to search architectures for Convolutional Neural Networks (CNN) and Graph Convolutional Networks (GCN). Extensive experiments show that SGAS is able to find state-of-the-art architectures for tasks such as image classification, point cloud classification and node classification in protein-protein interaction graphs with minimal computational cost.

We present a novel approach that converts partial and noisy RGB-D scans into high-quality 3D scene reconstructions by inferring unobserved scene geometry. Our approach is fully self-supervised and can hence be trained solely on incomplete, real-world scans. To achieve, self-supervision, we remove frames from a given (incomplete) 3D scan in order to make it even more incomplete; self-supervision is then formulated by correlating the two levels of partialness of the same scan while masking out regions that have never been observed. Through generalization across a large training set, we can then predict 3D scene completions even without seeing any 3D scan of entirely complete geometry. Combined with a new 3D sparse generative convolutional neural network architecture, our method is able to predict highly detailed surfaces in a coarse-to-fine hierarchical fashion that outperform existing state-of-the-art methods by a significant margin in terms of reconstruction quality.

Face image quality is an important factor to enable high-performance face recognition systems. Face quality assessment aims at estimating the suitability of a face image for the purpose of recognition. Previous work proposed supervised solutions that require artificially or human labelled quality values. However, both labelling mechanisms are error prone as they do not rely on a clear definition of quality and may not know the best characteristics for the utilized face recognition system. Avoiding the use of inaccurate quality labels, we proposed a novel concept to measure face quality based on an arbitrary face recognition model. By determining the embedding variations generated from random subnetworks of a face model, the robustness of a sample representation and thus, its quality is estimated. The experiments are conducted in a cross-database evaluation setting on three publicly available databases. We compare our proposed solution on two face embeddings against six state-of-the-art approaches from academia and industry. The results show that our unsupervised solution outperforms all other approaches in the majority of the investigated scenarios. In contrast to previous works, the proposed solution shows a stable performance over all scenarios. Utilizing the deployed face recognition model for our face quality assessment methodology avoids the training phase completely and further outperforms all baseline approaches by a large margin. Our solution can be easily integrated into current face recognition systems, and can be modified to other tasks beyond face recognition.

Scene text recognition is a hot research topic in computer vision. Recently, many recognition methods based on the encoder-decoder framework have been proposed, and they can handle scene texts of perspective distortion and curve shape. Nevertheless, they still face lots of challenges like image blur, uneven illumination, and incomplete characters. We argue that most encoder-decoder methods are based on local visual features without explicit global semantic information. In this work, we propose a semantics enhanced encoder-decoder framework to robustly recognize low-quality scene texts. The semantic information is used both in the encoder module for supervision and in the decoder module for initializing. In particular, the state-of-the-art ASTER method is integrated into the proposed framework as an exemplar. Extensive experiments demonstrate that the proposed framework is more robust for low-quality text images, and achieves state-of-the-art results on several benchmark datasets. The source code will be available.

Monocular depth estimation is an ill-posed problem, and as such critically relies on scene priors and semantics. Due to its complexity, we propose a deep neural network model based on a semantic divide-and-conquer approach. Our model decomposes a scene into semantic segments, such as object instances and background stuff classes, and then predicts a scale and shift invariant depth map for each semantic segment in a canonical space. Semantic segments of the same category share the same depth decoder, so the global depth prediction task is decomposed into a series of category-specific ones, which are simpler to learn and easier to generalize to new scene types. Finally, our model stitches each local depth segment by predicting its scale and shift based on the global context of the image. The model is trained end-to-end using a multi-task loss for panoptic segmentation and depth prediction, and is therefore able to leverage large-scale panoptic segmentation datasets to boost its semantic understanding. We validate the effectiveness of our approach and show state-of-the-art performance on three benchmark datasets.

Temporal action segmentation is a topic of increasing interest, however, annotating each frame in a video is cumbersome and costly. Weakly supervised approaches therefore aim at learning temporal action segmentation from videos that are only weakly labeled. In this work, we assume that for each training video only the list of actions is given that occur in the video, but not when, how often, and in which order they occur. In order to address this task, we propose an approach that can be trained end-to-end on such data. The approach divides the video into smaller temporal regions and predicts for each region the action label and its length. In addition, the network estimates the action labels for each frame. By measuring how consistent the frame-wise predictions are with respect to the temporal regions and the annotated action labels, the network learns to divide a video into class-consistent regions. We evaluate our approach on three datasets where the approach achieves state-of-the-art results.

The problem of counterfactual visual explanations is considered. A new family of discriminant explanations is introduced. These produce heatmaps that attribute high scores to image regions informative of a classifier prediction but not of a counter class. They connect attributive explanations, which are based on a single heat map, to counterfactual explanations, which account for both predicted class and counter class. The latter are shown to be computable by combination of two discriminant explanations, with reversed class pairs. It is argued that self-awareness, namely the ability to produce classification confidence scores, is important for the computation of discriminant explanations, which seek to identify regions where it is easy to discriminate between prediction and counter class. This suggests the computation of discriminant explanations by the combination of three attribution maps. The resulting counterfactual explanations are optimization free and thus much faster than previous methods. To address the difficulty of their evaluation, a proxy task and set of quantitative metrics are also proposed. Experiments under this protocol show that the proposed counterfactual explanations outperform the state of the art while achieving speeds much faster, for popular networks. In a human-learning machine teaching experiment, they are also shown to improve mean student accuracy from chance level to 95%.

Scene Text Recognition (STR), the task of recognizing text against complex image backgrounds, is an active area of research. Current state-of-the-art (SOTA) methods still struggle to recognize text written in arbitrary shapes. In this paper, we introduce a novel architecture for STR, named Selective Context ATtentional Text Recognizer (SCATTER). SCATTER utilizes a stacked block architecture with intermediate supervision during training, that paves the way to successfully train a deep BiLSTM encoder, thus improving the encoding of contextual dependencies. Decoding is done using a two-step 1D attention mechanism. The first attention step re-weights visual features from a CNN backbone together with contextual features computed by a BiLSTM layer. The second attention step, similar to previous papers, treats the features as a sequence and attends to the intra-sequence relationships. Experiments show that the proposed approach surpasses SOTA performance on irregular text recognition benchmarks by 3.7% on average.

Deep learning-based single image super-resolution (SISR) methods face various challenges when applied to 3D medical volumetric data (i.e., CT and MR images) due to the high memory cost and anisotropic resolution, which adversely affect their performance. Furthermore, mainstream SISR methods are designed to work over specific upsampling factors, which makes them ineffective in clinical practice. In this paper, we introduce a Spatially Aware Interpolation NeTwork (SAINT) for medical slice synthesis to alleviate the memory constraint that volumetric data poses. Compared to other super-resolution methods, SAINT utilizes voxel spacing information to provide desirable levels of details, and allows for the upsampling factor to be determined on the fly. Our evaluations based on 853 CT scans from four datasets that contain liver, colon, hepatic vessels, and kidneys show that SAINT consistently outperforms other SISR methods in terms of medical slice synthesis quality, while using only a single model to deal with different upsampling factors

We propose a sequential variational autoencoder to learn disentangled representations of sequential data (e.g., videos and audios) under self-supervision. Specifically, we exploit the benefits of some readily accessible supervision signals from input data itself or some off-the-shelf functional models and accordingly design auxiliary tasks for our model to utilize these signals. With the supervision of the signals, our model can easily disentangle the representation of an input sequence into static factors and dynamic factors (i.e., time-invariant and time-varying parts). Comprehensive experiments across videos and audios verify the effectiveness of our model on representation disentanglement and generation of sequential data, and demonstrate that, our model with self-supervision performs comparable to, if not better than, the fully-supervised model with ground truth labels, and outperforms state-of-the-art unsupervised models by a large margin.

Convolutional neural networks (CNNs) designed for low-dimensional regular grids will unfortunately lead to non-optimal solutions for analyzing spherical images, due to their different geometrical properties from planar images. In this paper, we generalize the grid-based CNNs to a non-Euclidean space by taking into account the geometry of spherical surfaces and propose a Spherical Graph Convolutional Network (SGCN) to encode rotation equivariant representations. Specifically, we propose a spherical graph construction criterion showing that a graph needs to be regular by evenly covering the spherical surfaces in order to design a rotation equivariant graph convolutional layer. For the practical case where the perfectly regular graph does not exist, we design two quantitative measures to evaluate the degree of irregularity for a spherical graph. The Geodesic ICOsahedral Pixelation (GICOPix) is adopted to construct spherical graphs with the minimum degree of irregularity compared to the current popular pixelation schemes. In addition, we design a hierarchical pooling layer to keep the rotation-equivariance, followed by a transition layer to enforce the invariance to the rotations for spherical image classification. We evaluate the proposed graph convolutional layers with different pixelations schemes in terms of equivariance errors. We also assess the effectiveness of the proposed SGCN in fulfilling rotation-invariance by the invariance error of the transition layers and recognizing the spherical images and 3D objects.

In this paper, we consider the low-bit quantization problem of face recognition (FR) under the open-set protocol. Different from well explored low-bit quantization on closed-set image classification task, the open-set task is more sensitive to quantization errors (QEs). We redefine the QEs in angular space and disentangle it into class error and individual error. These two parts correspond to inter-class separability and intra-class compactness, respectively. Instead of eliminating the entire QEs, we propose the rotation consistent margin (RCM) loss to minimize the individual error, which is more essential to feature discriminative power. Extensive experiments on popular benchmark datasets such as MegaFace Challenge, Youtube Faces (YTF), Labeled Face in the Wild (LFW) and IJB-C show the superiority of proposed loss in low-bit FR quantization tasks.

Though face rotation has achieved rapid progress in recent years, the lack of high-quality paired training data remains a great hurdle for existing methods. The current generative models heavily rely on datasets with multi-view images of the same person. Thus, their generated results are restricted by the scale and domain of the data source. To overcome these challenges, we propose a novel unsupervised framework that can synthesize photo-realistic rotated faces using only single-view image collections in the wild. Our key insight is that rotating faces in the 3D space back and forth, and re-rendering them to the 2D plane can serve as a strong self-supervision. We leverage the recent advances in 3D face modeling and high-resolution GAN to constitute our building blocks. Since the 3D rotation-and-render on faces can be applied to arbitrary angles without losing details, our approach is extremely suitable for in-the-wild scenarios (i.e. no paired data are available), where existing methods fall short. Extensive experiments demonstrate that our approach has superior synthesis quality as well as identity preservation over the state-of-the-art methods, across a wide range of poses and domains. Furthermore, we validate that our rotate-and-render framework naturally can act as an effective data augmentation engine for boosting modern face recognition systems even on strong baseline models

Deep Neural Networks are vulnerable to adversarial samples, which can fool classifiers by adding small perturbations onto the original image. Since the pioneering optimization-based adversarial attack method, many following methods have been proposed in the past several years. However most of these methods add perturbations in a "pixel-wise" and "global" way. Firstly, because of the contradiction between the local smoothness of natural images and the noisy property of these adversarial perturbations, this "pixel-wise" way makes these methods not robust to image processing based defense methods and steganalysis based detection methods. Secondly, we find adding perturbations to the background is less useful than to the salient object, thus the "global" way is also not optimal. Based on these two considerations, we propose the first robust superpixel-guided attentional adversarial attack method. Specifically, the adversarial perturbations are only added to the salient regions and guaranteed to be same within each superpixel. Through extensive experiments, we demonstrate our method can preserve the attack ability even in this highly constrained modification space. More importantly, compared to existing methods, it is significantly more robust to image processing based defense and steganalysis based detection.

In this paper, we propose a novel and efficient reference feature extraction module referred to as the Similarity Search and Extraction Network (SSEN) for reference-based super-resolution (RefSR) tasks. The proposed module extracts aligned relevant features from a reference image to increase the performance over single image super-resolution (SISR) methods. In contrast to conventional algorithms which utilize brute-force searches or optical flow estimations, the proposed algorithm is end-to-end trainable without any additional supervision or heavy computation, predicting the best match with a single network forward operation. Moreover, the proposed module is aware of not only the best matching position but also the relevancy of the best match. This makes our algorithm substantially robust when irrelevant reference images are given, overcoming the major cause of the performance degradation when using existing RefSR methods. Furthermore, our module can be utilized for self-similarity SR if no reference image is available. Experimental results demonstrate the superior performance of the proposed algorithm compared to previous works both quantitatively and qualitatively.

Various factors like occlusions, backgrounds, etc., would lead to misaligned detected bounding boxes , e.g., ones covering only portions of human body. This issue is common but overlooked by previous person search works. To alleviate this issue, this paper proposes an Align-to-Part Network (APNet) for person detection and re-Identification (reID). APNet refines detected bounding boxes to cover the estimated holistic body regions, from which discriminative part features can be extracted and aligned. Aligned part features naturally formulate reID as a partial feature matching procedure, where valid part features are selected for similarity computation, while part features on occluded or noisy regions are discarded. This design enhances the robustness of person search to real-world challenges with marginal computation overhead. This paper also contributes a Large-Scale dataset for Person Search in the wild (LSPS), which is by far the largest and the most challenging dataset for person search. Experiments show that APNet brings considerable performance improvement on LSPS. Meanwhile, it achieves competitive performance on existing person search benchmarks like CUHK-SYSU and PRW.

Detecting partially occluded objects is a difficult task. Our experimental results show that deep learning approaches, such as Faster R-CNN, are not robust at object detection under occlusion. Compositional convolutional neural networks (CompositionalNets) have been shown to be robust at classifying occluded objects by explicitly representing the object as a composition of parts. In this work, we propose to overcome two limitations of CompositionalNets which will enable them to detect partially occluded objects: 1) CompositionalNets, as well as other DCNN architectures, do not explicitly separate the representation of the context from the object itself. Under strong object occlusion, the influence of the context is amplified which can have severe negative effects for detection at test time. In order to overcome this, we propose to segment the context during training via bounding box annotations. We then use the segmentation to learn a context-aware compositionalNet that disentangles the representation of the context and the object. 2) We extend the part-based voting scheme in CompositionalNets to vote for the corners of the object's bounding box, which enables the model to reliably estimate bounding boxes for partially occluded objects. Our extensive experiments show that our proposed model can detect objects robustly, increasing the detection performance of strongly occluded vehicles from PASCAL3D+ and MS-COCO by 41% and 35% respectively in absolute performance relative to Faster R-CNN.