ECCV 2020 论文列表

Most state-of-the-art object detection systems follow an anchor-based diagram. Anchor boxes are densely proposed over the images and the network is trained to predict the boxes position offset as well as the classification confidence. Existing systems pre-define anchor box shapes and sizes and ad-hoc heuristic adjustments are used to define the anchor configurations. However, this might be sub-optimal or even wrong when a new dataset or a new model is adopted. In this paper, we study the problem of automatically optimizing anchor boxes for object detection. We first demonstrate that the number of anchors, anchor scales and ratios are crucial factors for a reliable object detection system. By carefully analyzing the existing bounding box patterns on the feature hierarchy, we design a flexible and tight hyper-parameter space for anchor configurations. Then we propose a novel hyper-parameter optimization method named AABO to determine more appropriate anchor boxes for a certain dataset, in which Bayesian Optimization and subsampling method are combined to achieve precise and efficient anchor configuration optimization. Experiments demonstrate the effectiveness of our proposed method on different detectors and datasets, e.g. achieving around 2.4% mAP improvement on COCO, 1.6% on ADE and 1.5% on VG, and the optimal anchors can bring 1.4% ∼ 2.4% mAP improvement on SOTA detectors by only optimizing anchor configurations, e.g. boosting Mask RCNN from 40.3% to 42.3%, and HTC detector from 46.8% to 48.2%."

Recently, substantial research efforts in Deep Metric Learning (DML) focused on designing complex pairwise-distance losses, which require convoluted schemes to ease optimization, such as sample mining or pair weighting. The standard cross-entropy loss for classification has been largely overlooked in DML. On the surface, the cross-entropy may seem unrelated and irrelevant to metric learning as it does not explicitly involve pairwise distances. However, we provide a theoretical analysis that links the cross-entropy to several well-known and recent pairwise losses. Our connections are drawn from two different perspectives: one based on an explicit optimization insight; the other on discriminative and generative views of the mutual information between the labels and the learned features. First, we explicitly demonstrate that the cross-entropy is an upper bound on a new pairwise loss, which has a structure similar to various pairwise losses: it minimizes intra-class distances while maximizing inter-class distances. As a result, minimizing the cross-entropy can be seen as an approximate bound-optimization (or Majorize-Minimize) algorithm for minimizing this pairwise loss. Second, we show that, more generally,minimizing the cross-entropy is actually equivalent to maximizing the mutual information, to which we connect several well-known pairwise losses. Furthermore, we show that various standard pairwise losses can be explicitly related to one another via bound relationships. Our findings indicate that the cross-entropy represents a proxy for maximizing the mutual information – as pairwise losses do – without the need for convoluted sample-mining heuristics. Our experiments†over four standard DML benchmarks strongly support our findings. We obtain state-of-the-art results, outperforming recent and complex DML methods."

We introduce UniLoss, a unified framework to generate surrogate losses for training deep networks with gradient descent, reducing the amount of manual design of task-specific surrogate losses. Our key observation is that in many cases, evaluating a model with a performance metric on a batch of examples can be refactored into four steps: from input to real-valued scores, from scores to comparisons of pairs of scores, from comparisons to binary variables, and from binary variables to the final performance metric. Using this refactoring we generate differentiable approximations for each non-differentiable step through interpolation. Using UniLoss, we can optimize for different tasks and metrics using one unified framework, achieving comparable performance compared with task-specific losses. We validate the effectiveness of UniLoss on three tasks and four datasets. Code is available at https://github.com/princeton-vl/uniloss."

In film making, shot has a profound influence on how the story is delivered and how the audiences are echoed. As different scale and movement types of shots can express different emotions and contents, recognizing shots and their attributes is important to the understanding of movies as well as general videos. Classifying shot type is challenging due to the additional information required beyond the video content, such as the spatial composition of a frame and the video camera movement. To address these issues, we propose a learning framework Subject Guidance Network (SGNet) for shot type recognition. SGNet separates the subject and background of a shot into two streams, serving as maps to guide scale and movement type classification respectively. To facilitate shot type analysis and model evaluations, we build a large-scale dataset MovieShots, which contains 46K shots from 7K movie trailers with annotations of their scale and movement types. Experiments show that our framework is able to recognize these two attributes of shot accurately, outperforming all the previous methods. "

Existing RGB-D salient object detection (SOD) approaches concentrate on the cross-modal fusion between the RGB stream and the depth stream. They do not deeply explore the effect of the depth map itself. In this work, we design a single stream network to directly use the depth map to guide early fusion and middle fusion between RGB and depth, which saves the feature encoder of the depth stream and achieves a lightweight and real-time model. We tactfully utilize depth information from two perspectives: (1) Overcoming the incompatibility problem caused by the great difference between modalities, we build a single stream encoder to achieve the early fusion, which can take full advantage of ImageNet pre-trained backbone model to extract rich and discriminative features. (2) We design a novel depth-enhanced dual attention module (DEDA) to efficiently provide the fore-/back-ground branches with the spatially filtered features, which enables the decoder to optimally perform the middle fusion. Besides, we put forward a pyramidally attended feature extraction module (PAFE) to accurately localize the objects of different scales. Extensive experiments demonstrate that the proposed model performs favorably against most state-of-the-art methods under different evaluation metrics. Furthermore, this model is 55.5\% lighter than the current lightest model and runs at a real-time speed of 32 FPS when processing a $384 imes 384$ image. "

Deep metric learning (DML) has received much attention in deep learning due to its wide applications in computer vision. Previous studies have focused on designing complicated losses and hard example mining methods, which are mostly heuristic and lack of theoretical understanding. In this paper, we cast DML as a simple pairwise binary classification problem that classifies a pair of examples as similar or dissimilar. It identifies the most critical issue in this problem---imbalanced data pairs. To tackle this issue, we propose a simple and effective framework to sample pairs in a batch of data for updating the model. The key to this framework is to define a robust loss for all pairs over a mini-batch of data, which is formulated by distributionally robust optimization. The flexibility in constructing the {\it uncertainty decision set} of the dual variable allows us to recover state-of-the-art complicated losses and also to induce novel variants. Empirical studies on several benchmark data sets demonstrate that our simple and effective method outperforms the state-of-the-art results."

The human visual system is remarkably robust against a wide range of naturally occurring variations and corruptions like rain or snow. In contrast, the performance of modern image recognition models strongly degrades when evaluated on previously unseen corruptions. Here, we demonstrate that a simple but properly tuned training with additive Gaussian and Speckle noise generalizes surprisingly well to unseen corruptions, easily reaching the state of the art on the corruption benchmark ImageNet-C (with ResNet50) and on MNIST-C. We build on top of these strong baseline results and show that an adversarial training of the recognition model against locally correlated worst-case noise distributions leads to an additional increase in performance. This regularization can be combined with previously proposed defense methods for further improvement."

In this work, we address the problem of synthesizing human actions from novel views. Given an input video of an actor performing some action, we aim to synthesize a video with the same action performed from a novel view with the help of an appearance prior. We propose an end-to-end deep network to solve this problem. The proposed network utilizes the change in viewpoint to transform the action from the input view to the novel view in feature space. The transformed action is integrated with the target appearance using the proposed recurrent transformer network, which provides a transformed appearance for each time-step in the action sequence. The recurrent transformer network utilize action key-points which are determined in an unsupervised approach using the encoded action features. We also propose a hierarchical structure for the recurrent transformation which further improves the performance. We demonstrate the effectiveness of the proposed method through extensive experiments conducted on a large-scale multi-view action recognition NTU-RGB+D dataset. In addition, we show that the proposed method can transform the action to a novel viewpoint with an entirely different scene or actor. The code is publicly available at https://github.com/schatzkara/cross-view-video."

Radon transform is a popular mathematical tool for shape analysis. However, it cannot handle affine deformation. Although its extended version, trace transform, allow us to construct affine invariants, they are less informative and computational expensive due to the loss of spatial relationship between trace lines and the extensive repeated calculation of transform. To address this issue, a novel line integral transform is proposed. We first use binding line pairs that have the desirable property of affine preserving as a reference frame to rewrite the diametrical dimension parameters of the lines in a relative manner which make them independent on affine transform. Along polar angle dimension of the line parameters, a moment-based normalization is then conducted to degrade the affine transform to similarity transform which can be easily normalized by Fourier transform. The proposed transform is not only invariant to affine transform, but also preserves the spatial relationship between line integrals which make it very informative. Another advantage of the proposed transform is that it is more efficient than the trace transform. Conducting it one time can allow us to achieve a 2D matrix of affine invariants. While conducting the trace transform once only generates a single feature and multiple trace transforms of different functionals are needed to derive more to make the descriptors informative. The effectiveness of the proposed transform has been validated on two types of standard shape test cases, affinely distorted contour shape dataset and region shape dataset, respectively."

Deep metric learning papers from the past four years have consistently claimed great advances in accuracy, often more than doubling the performance of decade-old methods. In this paper, we take a closer look at the field to see if this is actually true. We find flaws in the experimental setup of these papers, and propose a new way to evaluate metric learning algorithms. Finally, we present experimental results that show that the improvements over time have been marginal at best."

We introduce a large-scale annotated mechanical components benchmark for classification and retrieval tasks named MechanicalComponents Benchmark (MCB): a large-scale dataset of 3D objects of mechanical components. The dataset enables data-driven feature learn-ing for mechanical components. Exploring the shape descriptor for mechanical components is essential to computer vision and manufacturing applications. However, not much attention has been given on creating an-notated mechanical components datasets on a large-scale. This is because acquiring 3D models is challenging and annotating mechanical components requires engineering knowledge. Our main contributions are the creation of a large-scale annotated mechanical component benchmark, defining hierarchy taxonomy of mechanical components, and benchmark-ing the effectiveness of deep learning shape classifiers on the mechanical components. We created an annotated dataset and benchmarked seven state-of-the-art deep learning classification methods in three categories, namely: (1) point clouds, (2) volumetric representation in voxel grids, and (3) view-based representation."

Retrieval networks are essential for searching and indexing. Compared to classification networks, attention visualization for retrieval networks is hardly studied. We formulate attention visualization as a constrained optimization problem. We leverage the unit L2-Norm constraint as an attention filter (L2-CAF) to localize attention in both classification and retrieval networks. Unlike recent literature, our approach requires neither architectural changes nor fine-tuning. Thus, a pre-trained network's performance is never undermined L2-CAF is quantitatively evaluated using weakly supervised object localization. State-of-the-art results are achieved on classification networks. For retrieval networks, significant improvement margins are achieved over a Grad-CAM baseline. Qualitative evaluation demonstrates how the L2-CAF visualizes attention per frame for a recurrent retrieval network. Further ablation studies highlight the computational cost of our approach and compare L2-CAF with other feasible alternatives. Code available at https://bit.ly/3iDBLFv"

This work proposes a novel Graph-based neural ArchiTecture Encoding Scheme, a.k.a. GATES, to improve the predictor-based neural architecture search. Specifically, different from existing graph-based schemes, GATES models the operations as the transformation of the propagating information, which mimics the actual data processing of neural architecture. GATES is a more reasonable modeling of the neural architectures, and can encode architectures from both the ""operation on node"" and ""operation on edge"" cell search spaces consistently. Experimental results on various search spaces confirm GATES's effectiveness in improving the performance predictor. Furthermore, equipped with the improved performance predictor, the sample efficiency of the predictor-based neural architecture search (NAS) flow is boosted."

We present Generalized Histogram Thresholding (GHT), a simple, fast, and effective technique for histogram-based image thresholding. GHT works by performing approximate maximum a posteriori estimation of a mixture of Gaussians with appropriate priors. We demonstrate that GHT subsumes three classic thresholding techniques as special cases: Otsu's method, Minimum Error Thresholding (MET), and weighted percentile thresholding. GHT thereby enables the continuous interpolation between those three algorithms, which allows thresholding accuracy to be improved significantly. GHT also provides a clarifying interpretation of the common practice of coarsening a histogram's bin width during thresholding. We show that GHT outperforms or matches the performance of all algorithms on a recent challenge for handwritten document image binarization (including deep neural networks trained to produce per-pixel binarizations), and can be implemented in a dozen lines of code or as a trivial modification to Otsu's method or MET."

Video frame interpolation (VFI) aims at synthesizing new video frames in-between existing frames to generate smoother high frame rate videos. Current methods usually use the fixed pre-trained networks to generate interpolated-frames for different resolutions and scenes. However, the fixed pre-trained networks are difficult to be tailored for a variety of cases. Inspired by classical pyramid energy minimization optical flow algorithms, this paper proposes a recurrent residual pyramid network (RRPN) for video frame interpolation. In the proposed network, different pyramid levels share the same weights and base-network, named recurrent residual layer (RRL). In RRL, residual displacements between warped images are detected to gradually refine optical flows rather than directly predict the flows or frames. Owing to the flexible recurrent residual pyramid architecture, we can customize the number of pyramid levels, and make trade-offs between calculations and quality based on the application scenarios. Moreover, occlusion masks are also generated in this recurrent residual way to solve occlusion better. Finally, a refinement network is added to enhance the details for final output with contextual and edge information. Experimental results demonstrate that the RRPN is more flexible and efficient than current VFI networks but has fewer parameters. In particular, the RRPN, which avoid over-reliance on datasets and network structures, shows superior performance for large motion cases."

Dynamic Vision Sensors (DVSs) asynchronously stream events in correspondence of pixels subject to brightness changes. Differently from classic vision devices, they produce a sparse representation of the scene. Therefore, to apply standard computer vision algorithms, events need to be integrated into a frame or event-surface. This is usually attained through hand-crafted grids that reconstruct the frame using ad-hoc heuristics. In this paper, we propose Matrix-LSTM, a grid of Long Short-Term Memory (LSTM) cells that efficiently process events and learn end-to-end task-dependent event-surfaces. Compared to existing reconstruction approaches, our learned event-surface shows good flexibility and expressiveness on optical flow estimation on the MVSEC benchmark and it improves the state-of-the-art of event-based object classification on the N-Cars dataset."

The recently developed burst denoising approach, which reduces noise by using multiple frames captured in a short time, has demonstrated much better denoising performance than its single-frame counterparts. However, existing learning based burst denoising methods are limited by two factors. On one hand, most of the models are trained on video sequences with synthetic noise. When applied to real-world raw image sequences, visual artifacts often appear due to the different noise statistics. On the other hand, there lacks a real-world burst denoising benchmark of dynamic scenes because the generation of clean ground-truth is very difficult due to the presence of object motions. In this paper, a novel multi-frame CNN model is carefully designed, which decouples the learning of motion from the learning of noise statistics. Consequently, an alternating learning algorithm is developed to learn how to align adjacent frames from a synthetic noisy video dataset, and learn to adapt to the raw noise statistics from real-world noisy datasets of static scenes. Finally, the trained model can be applied to real-world dynamic sequences for burst denoising. Extensive experiments on both synthetic video datasets and real-world dynamic sequences demonstrate the leading burst denoising performance of our proposed method."

Autonomous agents must learn to collaborate. It is not scalable to develop a new centralized agent every time a task’s difficulty outpaces a single agent’s abilities. While multi-agent collaboration research has flourished in gridworld-like environments, relatively little work has considered visually rich domains. Addressing this, we introduce the novel task FurnMove in which agents work together to move a piece of furniture through a living room to a goal. Unlike existing tasks, FurnMove requires agents to coordinate at every timestep. We identify two challenges when training agents to complete FurnMove: existing decentralized action sampling procedures do not permit expressive joint action policies and, in tasks requiring close coordination, the number of failed actions dominates successful actions. To confront these challenges we introduce SYNC-policies (synchronize your actions coherently) and CORDIAL (coordination loss). Using SYNC-policies and CORDIAL, our agents achieve a 58% completion rate on FurnMove, an impressive absolute gain of 25 percentage points over competitive decentralized baselines. Our dataset, code, and pretrained models are available at https://unnat.github.io/cordial-sync ."

An approach for estimating the pose of a camera given a set of 3D points and their corresponding 2D image projections is presented. It formulates the problem as a non-linear quadratic program and identifies regions in the parameter space that contain unique minima with guarantees that at least one of them will be the global minimum. Each regional minimum is computed with a sequential quadratic programming scheme. These premises result in an algorithm that always determines the global minima of the perspective-n-point problem for any number of input correspondences, regardless of possible coplanar arrangements of the imaged 3D points. For its implementation, the algorithm merely requires ordinary operations available in any standard off-the-shelf linear algebra library. Comparative evaluation demonstrates that the algorithm achieves state-of-the-art results at a consistently low computational cost."