ECCV 2020 Papers

An interactive video object segmentation algorithm, which takes scribble annotations on query objects as input, is proposed in this paper. We develop a deep neural network, which consists of the annotation network (A-Net) and the transfer network (T-Net). First, given user scribbles on a frame, A-Net yields a segmentation result based on the encoder-decoder architecture. Second, T-Net transfers the segmentation result bidirectionally to the other frames, by employing the global and local transfer modules. The global transfer module conveys the segmentation information in an annotated frame to a target frame, while the local transfer module propagates the segmentation information in a temporally adjacent frame to the target frame. By applying A-Net and T-Net alternately, a user can obtain desired segmentation results with minimal efforts. We train the entire network in two stages, by emulating user scribbles and employing an auxiliary loss. Experimental results demonstrate that the proposed interactive video object segmentation algorithm outperforms the state-of-the-art conventional algorithms."

Interactive image restoration aims to generate restored images by adjusting a controlling coefficient which determines the restoration level. Previous works are restricted in modulating image with a single coefficient. However, real images always contain multiple types of degradation, which cannot be well determined by one coefficient. To make a step forward, this paper presents a new problem setup, called multi-dimension (MD) modulation, which aims at modulating output effects across multiple degradation types and levels. Compared with the previous single-dimension (SD) modulation, the MD is setup to handle multiple degradations adaptively and relief unbalanced learning problem in different degradations. We also propose a deep architecture - CResMD with newly introduced controllable residual connections for multi-dimension modulation. Specifically, we add a controlling variable on the conventional residual connection to allow a weighted summation of input and residual. The values of these weights are generated by another condition network. We further propose a new data sampling strategy based on beta distribution to balance different degradation types and levels. With corrupted image and degradation information as inputs, the network can output the corresponding restored image. By tweaking the condition vector, users can control the output effects in MD space at test time. Extensive experiments demonstrate that the proposed CResMD achieve excellent performance on both SD and MD modulation tasks. Code is available at \url{https://github.com/hejingwenhejingwen/CResMD}."

Inferring detailed 3D geometry of the scene is crucial for robotics applications, simulation, and 3D content creation. However, such information is hard to obtain, and thus very few datasets support it. In this paper, we propose an interactive framework for annotating 3D object geometry from both point cloud data and RGB imagery. The key idea behind our approach is to exploit strong priors that humans have about the 3D world in order to interactively annotate complete 3D shapes. Our framework targets a wide pool of annotators, i.e. naive users without artistic or graphics expertise. In particular, we introduce two simple-to-use interaction modules. First, we make an automatic guess of the 3D shape and allow the user to provide feedback about large errors by drawing scribbles in desired 2D views. Next, we aim to correct minor errors, in which users drag and drop 3D mesh vertices, assisted by a neural interactive module implemented as a Graph Convolutional Network. Experimentally, we show that only a few user interactions are needed to produce good quality 3D shapes on popular benchmarks such as ShapeNet, Pix3D, and ScanNet. We implement our framework as a web service and conduct a user study, where we show that user annotated data using our method effectively facilitates real-world learning tasks. Our web service will be released. "

Analysis of hand-hand interactions is a crucial step towards better understanding human behavior. However, most researches in 3D hand pose estimation have focused on the isolated single hand case. Therefore, we firstly propose (1) a large-scale dataset, InterHand2.6M, and (2) a baseline network, InterNet, for 3D interacting hand pose estimation from a single RGB image. The proposed InterHand2.6M consists of extbf{2.6M labeled single and interacting hand frames} under various poses from multiple subjects. Our InterNet simultaneously performs 3D single and interacting hand pose estimation. In our experiments, we demonstrate big gains in 3D interacting hand pose estimation accuracy when leveraging the interacting hand data in InterHand2.6M. We also report the accuracy of InterNet on InterHand2.6M, which serves as a strong baseline for this new dataset. Finally, we show 3D interacting hand pose estimation results from general images. Our code and dataset are availableootnote{\url{https://mks0601.github.io/InterHand2.6M/}}."

Weakly supervised localization aims at finding target object regions using only image-level supervision. However, localization maps extracted from classification networks are often not accurate due to the lack of fine pixel-level supervision. In this paper, we propose to leverage pixel-level similarities across different objects for learning more accurate object locations in a complementary way. Particularly, two kinds of constraints are proposed to prompt the consistency of object features within the same categories. The first constraint is to learn the stochastic feature consistency among discriminative pixels that are randomly sampled from different images within a batch. The discriminative information embedded in one image can be leveraged to benefit its counterpart with inter-image communication. The second constraint is to learn the global consistency of object features throughout the entire dataset. We learn a feature center for each category and realize the global feature consistency by forcing the object features to approach class-specific centers. The global centers are actively updated with the training process. The two constraints can benefit each other to learn consistent pixel-level features within the same categories, and finally improve the quality of localization maps. We conduct extensive experiments on two popular benchmarks, i.e., ILSVRC and CUB-200-2011. Our method achieves the Top-1 localization error rate of 45.17% on the ILSVRC validation set, surpassing the current state-of-the-art method by a large margin. The code is available at https://github.com/xiaomengyc/I2C."

Although recent works have made significant progress in encoding meaningful context information for instance segmentation in 2D images, the works for 3D point cloud counterpart lag far behind. Conventional methods use radius search or other similar methods for aggregating local information. However, these methods are unaware of the instance context and fail to realize the boundary and geometric information of an instance, which are critical to separate adjacent objects. In this work, we study the influence of instance-aware knowledge by proposing an Instance-Aware Module (IAM). The proposed IAM learns discriminative instance embedding features in two-fold: (1) Instance contextual regions, covering the spatial extension of an instance, are implicitly learned and propagated in the decoding process. (2) Instance-dependent geometric knowledge is included in the embedding space, which is informative and critical to discriminate adjacent instances. Moreover, the proposed IAM is free from complex and time-consuming operations, showing superiority in both accuracy and efficiency over the previous methods. To validate the effectiveness of our proposed method, comprehensive experiments have been conducted on three popular benchmarks for instance segmentation: ScannetV2, S3DIS, and PartNet. The flexibility of our method allows it to handle both indoor scenes and CAD objects. We achieve state-of-the-art performance, outperforming previous methods substantially."

The divergence between labeled training data and unlabeled testing data is a significant challenge for recent deep learning models. Unsupervised domain adaptation (UDA) attempts to solve such a problem. Recent works show that self-training is a powerful approach to UDA. However, existing methods have difficulty in balancing scalability and performance. In this paper, we propose an instance adaptive self-training framework for UDA on the task of semantic segmentation. To effectively improve the quality of pseudo-labels, we develop a novel pseudo-label generation strategy with an instance adaptive selector. Besides, we propose the region-guided regularization to smooth the pseudo-label region and sharpen the non-pseudo-label region. Our method is so concise and efficient that it is easy to be generalized to other unsupervised domain adaptation methods. Experiments on `GTA5 to Cityscapes' and `SYNTHIA to Cityscapes' demonstrate the superior performance of our approach compared with the state-of-the-art methods."

In the recent quest for trustworthy neural networks, we present Spiking Neural Network (SNN) as a potential candidate for inherent robustness against adversarial attacks. In this work, we demonstrate that adversarial accuracy of SNNs under gradient-based attacks is higher than their non-spiking counterparts for CIFAR datasets on deep VGG and ResNet architectures, particularly in blackbox attack scenario. We at-tribute this robustness to two fundamental characteristics of SNNs and analyze their effects. First, we exhibit that input discretization introduced by the Poisson encoder improves adversarial robustness with reduced number of timesteps. Second, we quantify the amount of adversarial accuracy with increased leak rate in Leaky-Integrate-Fire (LIF) neurons. Our results suggest that SNNs trained with LIF neurons and smaller number of timesteps are more robust than the ones with IF (Integrate-Fire) neurons and larger number of timesteps. Also we overcome the bottleneck of creating gradient-based adversarial inputs in temporal domain by proposing a technique for crafting attacks from SNN"

Multi-camera systems are an important sensor platform for intelligent systems such as self-driving cars. Pattern-based calibration techniques can be used to calibrate the intrinsics of the cameras individually. However, extrinsic calibration of systems with little to no visual overlap between the cameras is a challenge. Given the camera intrinsics, infrastucture-based calibration techniques are able to estimate the extrinsics using 3D maps pre-built via SLAM or Structure-from-Motion. In this paper, we propose to fully calibrate a multi-camera system from scratch using an infrastructure-based approach. Assuming that the distortion is mainly radial, we introduce a two-stage approach. We first estimate the camera-rig extrinsics up to a single unknown translation component per camera. Next, we solve for both the intrinsic parameters and the missing translation components. Extensive experiments on multiple indoor and outdoor scenes with multiple multi-camera systems show that our calibration method achieves high accuracy and robustness. In particular, our approach is more robust than the naive approach of first estimating intrinsic parameters and pose per camera before refining the extrinsic parameters of the system. The implementation is available at https://github.com/youkely/InfrasCal."

The performance of multi-domain image-to-image translation has been significantly improved by recent progress in deep generative models. Existing approaches can use a unified model to achieve translations between all the visual domains. However, their outcomes are far from satisfying when there are large domain variations. In this paper, we reveal that improving the sample selection strategy is an effective solution. To select informative samples, we dynamically estimate sample importance during the training of Generative Adversarial Networks, presenting Informative Sample Mining Network. We theoretically analyze the relationship between the sample importance and the prediction of the global optimal discriminator. Then a practical importance estimation function for general conditions is derived. In addition, we propose a novel multi-stage sample training scheme to reduce sample hardness while preserving sample informativeness. Extensive experiments on a wide range of specific image-to-image translation tasks are conducted, and the results demonstrate our superiority over current state-of-the-art methods."

Real-time 3D object detection is crucial for autonomous cars. Achieving promising performance with high efficiency, voxel-based approaches have received considerable attention. However, previous methods model the input space with features extracted from equally divided sub-regions without considering that point cloud is generally non-uniformly distributed over the space. To address this issue, we propose a novel 3D object detection framework with dynamic information modeling. The proposed framework is designed in a coarse-to-fine manner. Coarse predictions are generated in the first stage via a voxel-based region proposal network. We introduce InfoFocus, which improves the coarse detections by adaptively refining features guided by the information of point cloud density. Experiments are conducted on the large-scale nuScenes 3D detection benchmark. Results show that our framework achieves the state-of-the-art performance with 31 FPS and improves our baseline significantly by 9.0% mAP on the nuScenes test set."

A major endeavor of computer vision is to represent, understand and extract structure from 3D data. Towards this goal, unsupervised learning is a powerful and necessary tool. Most current unsupervised methods for 3D shape analysis use datasets that are aligned, require objects to be reconstructed and suffer from deteriorated performance on downstream tasks. To solve these issues we propose to extend the InfoMax and contrastive learning principles on 3D shapes. We show that we can maximize the mutual information between 3D objects and their ""chunks"" to improve the representations in aligned datasets. Furthermore, we can achieve rotation invariance in SO(3) group by maximizing the mutual information between the 3D objects and their geometric transformed versions. Finally, we conduct several experiments such as clustering, transfer learning, shape retrieval, and achieve state of art results."

Convolutional neural networks (CNNs) have achieved superior accuracy in many visual related tasks. However, the inference process through intermediate layers is opaque, making it difficult to interpret such networks or develop trust in their operation. We propose to model the network hidden layers activity using probabilistic models. The activity patterns in layers of interest are modeled as Gaussian mixture models, and transition probabilities between clusters in consecutive modeled layers are estimated. Based on maximum-likelihood considerations, a subset of the nodes and paths relevant for network prediction are chosen, connected, and visualized as an inference graph. We show that such graphs are useful for understanding the general inference process of a class, as well as explaining decisions the network makes regarding specific images. In addition, the models provide an interesting observation regarding the highly local nature of column activities in top CNN layers."

We present inertial safety maps (ISM), a novel scene representation designed for fast detection of obstacles in scenarios involving camera or scene motion, such as robot navigation and human-robot interaction. ISM is a motion-centric representation that encodes both scene geometry and motion; different camera motion results in different ISMs for the same scene. We show that ISM can be estimated with a two-camera stereo setup without explicitly recovering scene depths, by measuring differential changes in disparity over time. We develop an active, single-shot structured light-based approach for robustly measuring ISM in challenging scenarios with textureless objects and complex geometries. The proposed approach is computationally light-weight, and can detect intricate obstacles (e.g., thin wire fences) by processing high-resolution images at high-speeds with limited computational resources. ISM can be readily integrated with depth and range maps as a complementary scene representation, potentially enabling high-speed navigation and robotic manipulation in extreme environments, with minimal device complexity."

Fitting 3D morphable models (3DMMs) on faces is a well-studied problem, motivated by various industrial and research applications. 3DMMs express a 3D facial shape as a linear sum of basis functions. The resulting shape, however, is a plausible face only when the basis coefficients take values within limited intervals. Methods based on unconstrained optimization address this issue with a weighted $ll_2$ penalty on coefficients; however, determining the weight of this penalty is difficult, and the existence of a unique weight that works universally is questionable. We propose a new formulation that does not require the tuning of any weight parameter. Specifically, we formulate 3DMM fitting as an inequality-constrained optimization problem, where the primary constraint is that basis coefficients should not exceed the interval that is learned when the 3DMM is constructed. We employ additional constraints to exploit sparse landmark detectors by enforcing the facial shape to be within the error bounds of a reliable detector. To enable operation ``in-the-wild'', we use a robust objective function, namely gradient correlation. Our approach performs comparably with deep learning (DL) methods on ``in-the-wild'' data, which have inexact ground truth, and better than DL methods on more controlled data with exact ground truth. Since our formulation does not require any learning, it enjoys a versatility that allows it to operate with multiple frames of arbitrary sizes. This study's results encourage further research on 3DMM fitting with inequality-constrained optimization methods, which remained unexplored compared to unconstrained methods."

Conditional GANs (cGANs) are widely used in translating an image from one category to another. Meaningful conditions on GANsprovide greater flexibility and control over the nature of the target domain synthetic data. Existing conditional GANs commonly encode target domain label information as hard-coded categorical vectors in the form of0s and 1s. The major drawbacks of such representations are inability to encode the high-order semantic information of target categories and their relative dependencies. We propose a novel end-to-end learning framework based on Graph Convolutional Networks to learn the attribute representations to condition the generator. The GAN losses, the discriminator and attribute classification loss, are fed back to the graph resulting in the synthetic images that are more natural and clearer with respect to the attributes generation. Moreover, prior-arts are mostly given priorities to condition on the generator side, not on the discriminator side of GANs.We apply the conditions on the discriminator side as well via multi-task learning. We enhanced four state-of-the-art cGANs architectures: Stargan, Stargan-JNT, AttGAN and STGAN. Our extensive qualitative and quantitative evaluations on challenging face attributes manipulation data set,CelebA, LFWA, and RaFD, show that the cGANs enhanced by our meth-ods outperform by a large margin, compared to their counter-parts and other conditioning methods, in terms of both target attributes recognition rates and quality measures such as PSNR and SSIM."

Recently, deep networks have achieved impressive semantic segmentation performance, in particular thanks to their use of larger contextual information. In this paper, we show that the resulting networks are sensitive not only to global adversarial attacks, where perturbations affect the entire input image, but also to indirect local attacks, where the perturbations are confined to a small image region that does not overlap with the area that the attacker aims to fool. To this end, we introduce an indirect attack strategy, namely adaptive local attacks, aiming to find the best image location to perturb, while preserving the labels at this location and producing a realistic-looking segmentation map. Furthermore, we propose attack detection techniques both at the global image level and to obtain a pixel-wise localization of the fooled regions. Our results are unsettling: Because they exploit a larger context, more accurate semantic segmentation networks are more sensitive to indirect local attacks. We believe that our comprehensive analysis will motivate the community to design architectures with contextual dependencies that do not trade off robustness for accuracy."

We propose a method to train a model so it can learn new classification tasks while improving with each task solved. This amounts to combining meta-learning with incremental learning. Different tasks can have disjoint classes, so one cannot directly align different classifiers as done in model distillation. On the other hand, simply aligning features shared by all classes does not allow the base model sufficient flexibility to evolve to solve new tasks. We therefore indirectly align features relative to a minimal set of anchor classes. Such indirect discriminant alignment (IDA) adapts a new model to old classes without the need to re-process old data, while leaving maximum flexibility for the model to adapt to new tasks. This process enables incrementally improving the model by processing multiple learning episodes, each representing a different learning task, even with few training examples. Experiments on few-shot learning benchmarks show that this incremental approach performs favorably compared to training the model with the entire dataset at once. "

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