ECCV 2020 论文列表

Few-shot object detection (FSOD) helps detectors adapt to unseen classes with few training instances, and is useful when manual annotation is time-consuming or data acquisition is limited. Unlike previous attempts that exploit few-shot classification techniques to facilitate FSOD, this work highlights the necessity of handling the problem of scale variations, which is challenging due to the unique sample distribution. To this end, we propose a Multi-scale Positive Sample Refinement (MPSR) approach to enrich object scales in FSOD. It generates multi-scale positive samples as object pyramids and refines the prediction at various scales. We demonstrate its advantage by integrating it as an auxiliary branch to the popular architecture of Faster R-CNN with FPN, delivering a strong FSOD solution. Several experiments are conducted on PASCAL VOC and MS COCO, and the proposed approach achieves state of the art results and significantly outperforms other counterparts, which shows its effectiveness."

An algorithm to combine multiple loss terms adaptively for training a monocular depth estimator is proposed in this work. We construct a loss function space containing tens of losses. Using more losses can improve inference capability without any additional complexity in the test phase. However, when many losses are used, some of them may be neglected during training. Also, since each loss decreases at a different speed, adaptive weighting is required to balance the contributions of the losses. To address these issues, we propose the loss rebalancing algorithm that initializes and rebalances the weight for each loss function adaptively in the course of training. Experimental results show that the proposed algorithm provides state-of-the-art depth estimation results on various datasets. Codes are available at https://github.com/jaehanlee-mcl/multi-loss-rebalancing-depth."

We investigate a new AI task --- Multi-Agent Interactive Question Answering --- where several agents explore the scene jointly in interactive environments to answer a question. To cooperate efficiently and answer accurately, agents must be well-organized to have balanced work division and share knowledge about the objects involved. We address this new problem in two stages: Multi-Agent 3D Reconstruction in Interactive Environments and Question Answering. Our proposed framework features multi-layer structural and semantic memories shared by all agents, as well as a question answering model built upon a 3D-CNN network to encode the scene memories. During the reconstruction, agents simultaneously explore and scan the scene with a clear division of work, organized by next viewpoints planning. We evaluate our framework on the IQuADv1 dataset and outperform the IQA baseline in a single-agent scenario. In multi-agent scenarios, our framework shows favorable speedups while remaining high accuracy."

Video super-resolution (VSR) aims to utilize multiple low-resolution frames to generate a high-resolution prediction for each frame. In this process, inter- and intra-frames are the key sources for exploiting temporal and spatial information. However, there are a couple of limitations for existing VSR methods. First, optical flow is often used to establish one-on-one temporal correspondences. But flow estimation itself is error-prone and hence largely affects the ultimate recovery result. Second, similar patterns existing in natural images are rarely exploited for the VSR task. Motivated by these findings, we propose a temporal multi-correspondence aggregation strategy to leverage most similar patches across frames, and also a cross-scale nonlocal-correspondence aggregation scheme to explore self-similarity of images across scales. Based on these two novel modules, we build an effective multi-correspondence aggregation network (MuCAN) for VSR. Our method achieves state-of-the-art results on multiple benchmark datasets. Extensive experiments justify the effectiveness of our method."

Recent years have seen remarkable advances in visual understanding. However, how to understand a story-based long video with artistic styles, e.g. movie, remains challenging. In this paper, we introduce MovieNet -- a holistic dataset for movie understanding. MovieNet contains 1,100 movies with a large amount of multi-modal data, e.g. trailers, photos, plot descriptions, etc. Besides, different aspects of manual annotations are provided in MovieNet, including 1.1M characters with bounding boxes and identities, 42K scene boundaries, 2.5K aligned description sentences, 65K tags of place and action, and 92K tags of cinematic style. To the best of our knowledge, MovieNet is the largest dataset with richest annotations for comprehensive movie understanding. Based on MovieNet, we set up several benchmarks for movie understanding from different angles. Extensive experiments are executed on these benchmarks to show the immeasurable value of MovieNet and the gap of current approaches towards comprehensive movie understanding. We believe that such a holistic dataset would promote the researches on story-based long video understanding and beyond. MovieNet will be published in compliance with regulations at https://movienet.github.io."

Motion plays a crucial role in understanding videos and most state-of-the-art neural models for video classification incorporate motion information typically using optical flows extracted by a separate off-the-shelf method. As the frame-by-frame optical flows require heavy computation, incorporating motion information has remained a major computational bottleneck for video understanding. In this work, we replace external and heavy computation of optical flows with internal and light-weight learning of motion features. We propose a trainable neural module, dubbed MotionSqueeze, for effective motion feature extraction. Inserted in the middle of any neural network, it learns to establish correspondences across frames and convert them into motion features, which are readily fed to the next downstream layer for better prediction. We demonstrate that the proposed method provides a significant gain on four standard benchmarks for action recognition with only a small amount of additional cost, outperforming the state of the art on Something-Something-V1&V2 datasets."

Deep neural networks are known to be susceptible to adversarial noise, which is tiny and imperceptible perturbation. Most of previous works on adversarial attack mainly focus on image models, while the vulnerability of video models is less explored. In this paper, we aim to attack video models by utilizing intrinsic movement pattern and regional relative motion among video frames. We propose an effective motion-excited sampler to obtain motion-aware noise prior, which we term as sparked prior. Our sparked prior underlines frame correlations and utilizes video dynamics via relative motion. By using the sparked prior in gradient estimation, we can successfully attack a variety of video classification models with fewer number of queries. Extensive experimental results on four benchmark datasets validate the efficacy of our proposed method. "

We propose a new loss function, called motion loss, for the problem of monocular 3D Human pose estimation from 2D pose. It introduces the task of reconstructing keypoint motion into supervision. In computing motion loss, a simple yet effective representation for keypoint motion, called multiscale motion encoding, is introduced. We design a new graph convolutional network architecture, U-shaped GCN (UGCN). It captures both short-term and long-term motion information to fully leverage the additional supervision from the motion loss. We experiment training UGCN with the motion loss on two large scale benchmarks: Human3.6M and MPI-INF-3DHP. Our model surpasses other state-of-the-art models by a large margin. It also demonstrates strong capacity in producing smooth 3D sequences and recovering keypoint motion."

Recent advances in image-based human pose estimation make it possible to capture 3D human motion from a single RGB video. However, the inherent depth ambiguity and self-occlusion in a single view prohibit the recovery of as high-quality motion as multi-view reconstruction. While multi-view videos are not common, the videos of a celebrity performing a specific action are usually abundant on the Internet. Even if these videos were recorded at different time instances, they would encode the same motion characteristics of the person. Therefore, we propose to capture human motion by jointly analyzing these Internet videos instead of using single videos separately. However, this new task poses many new challenges that cannot be addressed by existing methods, as the videos are unsynchronized, the camera viewpoints are unknown, the background scenes are different, and the human motions are not exactly the same among videos. To address these challenges, we propose a novel optimization-based framework and experimentally demonstrate its ability to recover much more precise and detailed motion from multiple videos, compared against monocular motion capture methods. "

Less Forgetting: A Generic Multi-classifier Paradigm for Incremental Learning","Overcoming catastrophic forgetting in neural networks is a long-standing and core research objective for incremental learning. Notable studies have shown regularization strategies enable the network to remember previously acquired knowledge devoid of heavy forgetting. Since those regularization strategies are mostly associated with classifier outputs, we propose a MUlti-Classifier (MUC) incremental learning paradigm that integrates an ensemble of auxiliary classifiers to estimate more effective regularization constraints. Additionally, we extend two common methods, focusing on parameter and activation regularization, from the conventional single-classifier paradigm to MUC. Our classifier ensemble promotes regularizing network parameters or activations when moving to learn the next task. Under the setting of task-agnostic evaluation, our experimental results on CIFAR-100 and Tiny ImageNet incremental benchmarks show that our method outperforms other baselines. Specifically, MUC obtains 3%-5% accuracy boost and 4%-5% decline of forgetting ratio, compared with MAS and LwF. Our code is available at https://github.com/Liuy8/MUC. "

We introduce a differentiable monotonicity prior, useful to express signals of monotonic tendency. An important natural signal of this tendency is the optical extinction coefficient, as a function of altitude in a cloud. Cloud droplets become larger as vapor condenses on them in an updraft. Reconstruction of the volumetric structure of clouds is important for climate research. Data for such reconstruction is multi-view images taken simultaneously of each cloud. This acquisition mode is expected by upcoming future spaceborne imagers. We achieve three-dimensional volumetric reconstruction through stochastic scattering tomography, which is based on optimization of a cost function. Part of the cost is the monotonicity prior, which helps to improve the reconstruction quality. The stochastic tomography is based on Monte-Carlo radiative transfer. It is formulated and implemented in a coarse-to-fine form, making it scalable to large fields."

We present the first approach to volumetric performance capture and novel-view rendering at real-time speed from monocular video, eliminating the need for expensive multi-view systems or cumbersome pre-acquisition of a personalized template model. Our system reconstructs a fully textured 3D human from each frame by leveraging Pixel-Aligned Implicit Function (PIFu). While PIFu achieves high-resolution reconstruction in a memory-efficient manner, its computationally expensive inference prevents us from deploying such a system for real-time applications. To this end, we propose a novel hierarchical surface localization algorithm and a direct rendering method without explicitly extracting surface meshes. By culling unnecessary regions for evaluation in a coarse-to-fine manner, we successfully accelerate the reconstruction by two orders of magnitude from the baseline without compromising the quality. Furthermore, we introduce an Online Hard Example Mining (OHEM) technique that effectively suppresses failure modes due to the rare occurrence of challenging examples. We adaptively update the sampling probability of the training data based on the current reconstruction accuracy, which effectively alleviates reconstruction artifacts. Our experiments and evaluations demonstrate the robustness of our system to various challenging angles, illuminations, poses, and clothing styles. We also show that our approach compares favorably with state-of-the-art monocular performance capture. Our proposed approach removes the need for multi-view studio settings and enables a consumer-accessible solution for volumetric capture."

To understand how people look, interact, or perform tasks, we need to quickly and accurately capture their 3D body, face, and hands together from an RGB image. Most existing methods focus only on parts of the body. A few recent approaches reconstruct full expressive 3D humans from images using 3D body models that include the face and hands. These methods are optimization-based and thus slow, prone to local optima, and require 2D keypoints as input. We address these limitations by introducing ExPose (EXpressive POse and Shape rEgression), which directly regresses the body, face, and hands, in SMPL-X format, from an RGB image. This is a hard problem due to the high dimensionality of the body and the lack of expressive training data. Additionally, hands and faces are much smaller than the body, occupying very few image pixels. This makes hand and face estimation hard when body images are downscaled for neural networks. We make three main contributions. First, we account for the lack of training data by curating a dataset of SMPL-X fits on in-the-wild images. Second, we observe that body estimation localizes the face and hands reasonably well. We introduce body-driven attention for face and hand regions in the original image to extract higher-resolution crops that are fed to dedicated refinement modules. Third, these modules exploit part-specific knowledge from existing face and hand-only datasets. ExPose estimates expressive 3D humans more accurately than existing optimization methods at a small fraction of the computational cost. Our data, model and code are available for research at https://expose.is.tue.mpg.de."

3D object detection from monocular images is an ill-posed problem due to the projective entanglement of depth and scale. To overcome this ambiguity, we present a novel self-supervised method for textured 3D shape reconstruction and pose estimation of rigid objects with the help of strong shape priors and 2D instance masks. Our method predicts the 3D location and meshes of each object in an image using differentiable rendering and a self-supervised objective derived from a pretrained monocular depth estimation network. We use the KITTI 3D object detection dataset to evaluate the accuracy of the method. Experiments demonstrate that we can effectively use noisy monocular depth and differentiable rendering as an alternative to expensive 3D ground-truth labels or LiDAR information."

Monocular 3D object detection is a challenging task due to unreliable depth, resulting in a distinct performance gap between monocular and LiDAR-based approaches. In this paper, we propose a novel domain adaptation based monocular 3D object detection framework named DA-3Ddet, which adapts the feature from unsound image-based pseudo-LiDAR domain to the accurate real LiDAR domain for performance boosting. In order to solve the overlooked problem of inconsistency between the foreground mask of pseudo and real LiDAR caused by inaccurately estimated depth, we also introduce a context-aware foreground segmentation module which helps to involve relevant points for foreground masking. Extensive experiments on KITTI dataset demonstrate that our simple yet effective framework outperforms other state-of-the-arts by a large margin."

Normalization layers play an important role in deep network training. As one of the most popular normalization techniques, batch normalization (BN) has shown its effectiveness in accelerating the model training speed and improving model generalization capability. The success of BN has been explained from different views, such as reducing internal covariate shift, allowing the use of large learning rate, smoothing optimization landscape, etc. To make a deeper understanding of BN, in this work we prove that BN actually introduces a certain level of noise into the sample mean and variance during the training process, while the noise level depends only on the batch size. Such a noise generation mechanism of BN regularizes the training process, and we present an explicit regularizer formulation of BN. Since the regularization strength of BN is determined by the batch size, a small batch size may cause the under-fitting problem, resulting in a less effective model. To reduce the dependency of BN on batch size, we propose a momentum BN (MBN) scheme by averaging the mean and variance of current mini-batch with the historical means and variances. With a dynamic momentum parameter, we can automatically control the noise level in the training process. As a result, MBN works very well even when the batch size is very small (e.g., 2), which is hard to achieve by traditional BN."

Surface registration plays a fundamental role in shape analysis and geometric processing. Generally, there are three criteria in evaluating a surface mapping result: diffeomorphism, small distortion and feature alignment. In order to fulfill these requirements, this work proposes a novel model of the space of point landmark constrained diffeomorphisms. Based on Teichm\""uller theory, this mapping space is generated by the Beltrami coefficients, which are infinitesimally Teichm\""uller equivalent to $0$. These Beltrami coefficients are the solutions to a linear equation group. By using this theoretic model, optimal registrations can be achieved by iterative optimization with linear constraints in the diffeomorphism space, such as harmonic maps and Teichm\""uller maps, which minimize different type of distortions. The theoretic model is rigorous and has practical value. Our experiment results demonstrate the efficiency and efficacy of the proposed method."

In this paper, we study the effects of windshield refraction for autonomous driving applications. These distortion effects are surprisingly large and can not be explained by traditional camera models. Instead of using a generalized camera approach, we propose a novel approach to jointly optimize a traditional camera model, and a mathematical representation of the windshield’s surface. First, using the laws of geometric optics, the refraction is modeled using a local spherical approximation to the windshield’s geometry. Next, a spline-based model is proposed as a refinement to better adapt to deviations from the ideal shape and manufacturing variations. By jointly optimizing refraction and camera parameters, the projection error can be significantly reduced. The proposed models are validated on real windshield observations and custom setups to compare recordings with and without windshield, with accurate laser scan measurements as 3D ground truth."

People often create art by following an artistic workflow involving multiple stages that inform the overall design. If an artist wishes to modify an earlier decision, significant work may be required to propagate this new decision forward to the final artwork. Motivated by the above observations, we propose a generative model that follows a given artistic workflow, enabling both multi-stage image generation as well as multi-stage image editing of an existing piece of art. Furthermore, for the editing scenario, we introduce an optimization process along with learning-based regularization to ensure the edited image produced by the model closely aligns with the originally provided image. Qualitative and quantitative results on three different artistic datasets demonstrate the effectiveness of the proposed framework on both image generation and editing tasks."