ECCV 2020 Papers

Learning discriminative powerful representations is a crucial step for machine learning systems. Introducing invariance against arbitrary nuisance or sensitive attributes while performing well on specific tasks is an important problem in representation learning. This is mostly approached by purging the sensitive information from learned representations. In this paper, we propose a novel disentanglement approach to invariant representation problem. We disentangle the meaningful and sensitive representations by enforcing orthogonality constraints as a proxy for independence. We explicitly enforce the meaningful representation to be agnostic to sensitive information by entropy maximization. The proposed approach is evaluated on five publicly available datasets and compared with state of the art methods for learning fairness and invariance achieving the state of the art performance on three datasets and comparable performance on the rest. Further, we perform an ablative study to evaluate the effect of each component."

Algorithmic decision making based on computer vision and machine learning technologies continue to permeate our lives. But issues related to biases of these models and the extent to which they treat certain segments of the population unfairly, have led to concern in the general public. It is now accepted that because of biases in the datasets we present to the models, a fairness-oblivious training will lead to unfair models. An interesting topic is the study of mechanisms via which the de novo design or training of the model can be informed by fairness measures. Here, we study mechanisms that impose fairness concurrently while training the model. While existing fairness based approaches in vision have largely relied on training adversarial modules together with the primary classification/regression task, in an effort to remove the influence of the protected attribute or variable, we show how ideas based on well-known optimization concepts can provide a simpler alternative. In our proposed scheme, imposing fairness just requires specifying the protected attribute and utilizing our optimization routine. We provide a detailed technical analysis and present experiments demonstrating that various fairness measures from the literature can be reliably imposed on a number of training tasks in vision in a manner that is interpretable."

Differentiable Architecture Search (DARTS) is now a widely disseminated weight-sharing neural architecture search method. However, it suffers from well-known performance collapse due to an inevitable aggregation of skip connections. In this paper, we first disclose that its root cause lies in an unfair advantage in exclusive competition. Through experiments, we show that if either of two conditions is broken, the collapse disappears. Thereby, we present a novel approach called Fair DARTS where the exclusive competition is relaxed to be collaborative. Specifically, we let each operation's architectural weight be independent of others. Yet there is still an important issue of discretization discrepancy. We then propose a zero-one loss to push architectural weights towards zero or one, which approximates an expected multi-hot solution. Our experiments are performed on two mainstream search spaces, and we derive new state-of-the-art results on CIFAR-10 and ImageNet."

State-of-the-art face super-resolution methods employ deep convolutional neural networks to learn a mapping between low- and high-resolution facial patterns by exploring local appearance knowledge. However, most of these methods do not well exploit facial structures and identity information, and struggle to deal with facial images that exhibit large pose variations. In this paper, we propose a novel face super-resolution method that explicitly incorporates 3D facial priors which grasp the sharp facial structures. Our work is the first to explore 3D morphable knowledge based on the fusion of parametric descriptions of face attributes (e.g., identity, facial expression, texture, illumination, and face pose). Furthermore, the priors can easily be incorporated into any networks and are extremely efficient in improving the performance and accelerating the convergence speed. Firstly, a 3D face rendering branch is set up to obtain 3D priors of salient facial structures and identity knowledge. Secondly, the Spatial Attention Module is used to better exploit this hierarchical information (i.e., intensity similarity, 3D facial structure, and identity content) for the super-resolution problem. Extensive experiments demonstrate that the proposed 3D priors achieve superior face super-resolution results over the state-of-the-arts."

Face anti-spoofing (FAS) plays a vital role in securing the face recognition systems from presentation attacks. Most existing FAS methods capture various cues (e.g., texture, depth and reflection) to distinguish the live faces from the spoofing faces. All these cues are based on the discrepancy among physical materials (e.g., skin, glass, paper and silicone). In this paper we rephrase face anti-spoofing as a material recognition problem and combine it with classical human material perception, intending to extract discriminative and robust features for FAS. To this end, we propose the Bilateral Convolutional Networks (BCN), which is able to capture intrinsic material-based patterns via aggregating multi-level bilateral macro- and micro- information. Furthermore, Multi-level Feature Refinement Module (MFRM) and multi-head supervision are utilized to learn more robust features. Comprehensive experiments are performed on six benchmark datasets, and the proposed method achieves superior performance on both intra- and cross-dataset testings. One highlight is that we achieve overall 11.3$\pm$9.5\% EER for cross-type testing in SiW-M dataset, which significantly outperforms previous results. We hope this work will facilitate future cooperation between FAS and material communities. "

Face anti-spoofing is crucial to the security of face recognition systems. Previous approaches focus on developing discriminative models based on the features extracted from images, which may be still entangled between spoof patterns and real persons. In this paper, motivated by the disentangled representation learning, we propose a novel perspective of face anti-spoofing that disentangles the liveness features and content features from images, and the liveness features is further used for classification. We also put forward a CNN architecture with the process of disentanglement and combination of low-level and high-level supervision to improve the generalization capabilities. We evaluate our method on public benchmark datasets and extensive experimental results demonstrate the effectiveness of our method against the state-of-the-art competitors. Finally, we further visualize some results to help understand the effect and advantage of disentanglement. "

Low-bit Deep Neural Networks (low-bit DNNs) have recently received significant attention for their high efficiency. However, low-bit DNNs are often difficult to optimize due to the the saddle points in loss surfaces. Here we introduce a novel feature-based knowledge transfer framework, which utilizes a 32-bit DNN to guide the training of a low-bit DNN via feature maps. It is challenge because feature maps from two branches lie in continuous and discrete space respectively, and such mismatch has not been handled properly by existing feature transfer frameworks. In this paper, we propose to directly transfer information-rich continuous-space feature to the low-bit branch. To alleviate the negative impacts brought by the feature quantizer during the transfer process, we make two branches interact via centered cosine distance rather than the widely-used p-norms. Extensive experiments are conducted on Cifar10/100 and ImageNet. Compared with low-bit models trained directly, the proposed framework brings 0.5% to 3.4% accuracy gains to three different quantization schemes. Besides, the proposed framework can also be combined with other techniques, e.g. logits transfer, for further enhacement."

We propose and study a method called FLOT that estimates scene flow on point clouds. We start the design of FLOT by noticing that scene flow estimation on point clouds reduces to estimating a permutation matrix in a perfect world. Inspired by recent works on graph matching, we build a method to find these correspondences by borrowing tools from optimal transport. Then, we relax the transport constraints to take into account real-world imperfections. The transport cost between two points is given by the pairwise similarity between deep features extracted by a neural network trained under full supervision using synthetic datasets. Our main finding is that FLOT can perform as well as the best existing methods on synthetic and real-world datasets while requiring much less parameters and without using multiscale analysis. Our second finding is that, on the training datasets considered, most of the performance can be explained by the learned transport cost. This yields a simpler method, FLOT$_0$, which is obtained using a particular choice of optimal transport parameters and performs nearly as well as FLOT."

Frequency aliasing in the digital capture of display screens leads to the moir´e pattern, appearing as stripe-shaped distortions in images. Efforts to demoir´eing have been made recently in a learning fashion due to the complexity and diversity of the pattern appearance. However, existing methods cannot satisfy the practical demand of demoir´eing on camera phone capturing more pixels than a full high definition (FHD) image, which poses additional challenges of wider pattern scale range and fine detail preservation. We propose the Full High Definition Demoir´eing Network (FHDe$^2$Net) to solve such problems. The framework consists of a global to local cascaded removal branch to eradicate multi-scale moir´e patterns and a frequency based high resolution content separation branch to retain fine details. We further collect an FHD moir´e image dataset as a new benchmark for training and evaluation. Comparison experiments and ablation studies have verified the effectiveness of the proposed framework and each functional module both quantitatively and qualitatively in practical application scenarios."

A complete 3D face reconstruction requires to explicitly model the eyeglasses on the face, which is less investigated in the literature. In this paper, we present an automatic system that recovers the 3D shape of eyeglasses from a single face image with an arbitrary head pose. To achieve this goal, our system first trains a neural network to jointly perform glasses landmark detection and segmentation, which carry the sparse and dense glasses information respectively for 3D glasses pose estimation and shape recovery. To solve the ambiguity in 2D to 3D reconstruction, our system fully explores the prior knowledge including the relative motion constraint between face and glasses and the planar and symmetric shape prior feature of glasses. From the qualitative and quantitative experiments, we see that our system reconstructs promising 3D shapes of eyeglasses for various poses."

For a certain object in an image, the relationship between its central region and the peripheral region is not well utilized in existing superpixel segmentation methods. In this work, we propose the concept of regional attribute, which indicates the location of a certain region in the object. Based on the regional attributes, we propose a novel superpixel method called Extract and Merge (EAM). In the extracting stage, we design square windows with a side length of a power of two, named power-window, to extract regional attributes by calculating boundary clearness of objects in the window. The larger windows are for the central regions and the smaller ones correspond to the peripheral regions. In the merging stage, power-windows are merged according to the defined attraction between them. Specifically, we build a graph model and propose an efficient method to make the large windows merge the small ones strategically, regarding power-windows as vertices and the adjacencies between them as edges. We demonstrate that our superpixels have fine boundaries and are superior to the respective state-of-the-art algorithms on multiple benchmarks."

Self-supervised representation learning has achieved impressive results in recent years, with experiments primarily coming on ImageNet or other similarly large internet imagery datasets. There has been little to no work with these methods on other smaller domains, such as satellite, textural, or biological imagery. We experiment with several popular methods on an unprecedented variety of domains. We discover, among other findings, that Rotation is the most semantically meaningful task, while much of the performance of Jigsaw is attributable to the nature of its induced distribution rather than semantic understanding. Additionally, there are several areas, such as fine-grain classification, where all tasks underperform. We quantitatively and qualitatively diagnose the reasons for these failures and successes via novel experiments studying pretext generalization, random labelings, and implicit dimensionality. Code and models are available at https://github.com/BramSW/Extending_SSRL_Across_Domains/."

VR telepresence consists of interacting with another human in a virtual space represented by an avatar. Today most avatars are cartoon-like, but soon the technology will allow video-realistic ones. This paper aims in this direction and presents Modular Codec Avatars (MCA), a method to generate hyper-realistic faces driven by the cameras in the VR headset. MCA extends traditional Codec Avatars (CA) by replacing the holistic models with a learned modular representation. It is important to note that traditional person-specific CAs are learned from few training samples, and typically lack robustness as well as limited expressiveness when transferring facial expressions. MCAs solve these issues by learning a modulated adaptive blending of different facial components as well as an exemplar-based latent alignment. We demonstrate that MCA achieves improved expressiveness and robustness w.r.t to CA in a variety of real-world datasets and practical scenarios. Finally, we showcase new applications in VR telepresence enabled by the proposed model."

While supervised techniques in re-identification are extremely effective, the need for large amounts of annotations makes them impractical for large camera networks. One-shot re-identification, which uses a singular labeled tracklet for each identity along with a pool of unlabeled tracklets, is a potential candidate towards reducing this labeling effort. Current one-shot re-identification methods function by modeling the inter-relationships amongst the labeled and the unlabeled data, but fail to fully exploit such relationships that exist within the pool of unlabeled data itself. In this paper, we propose a new framework named Temporal Consistency Progressive Learning, which uses temporal coherence as a novel self-supervised auxiliary task in the one-shot learning paradigm to capture such relationships amongst the unlabeled tracklets. Optimizing two new losses, which enforce consistency on a local and global scale, our framework can learn learn richer and more discriminative representations. Extensive experiments on two challenging video re-identification datasets - MARS and DukeMTMC-VideoReID - demonstrate that our proposed method is able to estimate the true labels of the unlabeled data more accurately by up to $8\%$, and obtain significantly better re-identification performance compared to the existing state-of-the-art techniques."

Learning a good image prior is a long-term goal for image restoration and manipulation. While existing methods like deep image prior (DIP) capture low-level image statistics, there are still gaps toward an image prior that captures rich image semantics including color, spatial coherence, textures, and high-level concepts. This work presents an effective way to exploit the image prior captured by a generative adversarial network (GAN) trained on large-scale natural images. As shown in Fig.1, the deep generative prior (DGP) provides compelling results to restore missing semantics, e.g., color, patch, resolution, of various degraded images. It also enables diverse image manipulation including random jittering, image morphing, and category transfer. Such highly flexible effects are made possible through relaxing the assumption of existing GAN-inversion methods, which tend to fix the generator. Notably, we allow the generator to be fine-tuned on-the-fly in a progressive manner regularized by feature distance obtained by the discriminator in GAN. We show that these easy-to-implement and practical changes help preserve the reconstruction to remain in the manifold of nature image, and thus lead to more precise and faithful reconstruction for real images. Code is at https://github.com/XingangPan/deep-generative-prior."

Visual relationship detection is fundamental for holistic image understanding. However, the localization and classification of (subject, predicate, object) triplets remain challenging tasks, due to the combinatorial explosion of possible relationships, their long-tailed distribution in natural images, and an expensive annotation process. This paper introduces a novel weakly-supervised method for visual relationship detection that relies on minimal image-level predicate labels. A graph neural network is trained to classify predicates in images from a graph representation of detected objects, implicitly encoding an inductive bias for pairwise relations. We then frame relationship detection as the explanation of such a predicate classifier, i.e. we obtain a complete relation by recovering the subject and object of a predicted predicate. We present results comparable to recent fully- and weakly-supervised methods on three diverse and challenging datasets: HICO-DET for human-object interaction, Visual Relationship Detection for generic object-to-object relations, and UnRel for unusual triplets; demonstrating robustness to non-comprehensive annotations and good few-shot generalization."

The black-box nature of deep neural networks (DNNs) makes it impossible to understand why a particular output is produced, creating demand for “Explainable AI”. In this paper, we show that statistical fault localization (SFL) techniques from software engineering deliver high quality explanations of the outputs of DNNs, where we define an explanation as a minimal subset of features sufficient for making the same decision as for the original input. We present an algorithm and a tool called DeepCover, which synthesizes a ranking of the features of the inputs using SFL and constructs explanations for the decisions of the DNN based on this ranking. We compare explanations produced by DeepCover with those of the state-of-the-art tools GradCAM, LIME, SHAP, RISE and Extremal and show that explanations generated by DeepCover are consistently better across a broad set of experiments. On a benchmark set with known ground truth, DeepCover achieves 76.7% accuracy, which is 6% better than the second best Extremal."

Explainable face recognition (XFR) is the problem of explaining the matches returned by a facial matcher, in order to provide insight into why a probe was matched with one identity over another. In this paper, we provide the first comprehensive benchmark and baseline evaluation for XFR. We define a new evaluation protocol called the ``inpainting game'', which is a curated set of 3648 triplets (probe, mate, nonmate) of 95 subjects, which differ by synthetically inpainting a chosen facial characteristic like the nose, eyebrows or mouth creating an inpainted nonmate. An XFR algorithm is tasked with generating a network attention map which best explains which regions in a probe image match with a mated image, and not with an inpainted nonmate for each triplet. This provides ground truth for quantifying what image regions contribute to face matching. Finally, we provide a comprehensive benchmark on this dataset comparing five state-of-the-art XFR algorithms on three facial matchers. This benchmark includes two new algorithms called subtree EBP and Density-based Input Sampling for Explanation (DISE) which outperform the state-of-the-art XFR by a wide margin."

Knowledge distillation is an effective tool to compress large pre-trained Convolutional Neural Networks (CNNs) or their ensembles into models applicable to mobile and embedded devices. The success of which mainly comes from two aspects: the designed student network and the exploited knowledge. However, current methods usually suffer from the low-capability of mobile-level student network and the unsatisfactory knowledge for distillation. In this paper, we propose a novel position-aware exclusivity to encourage large diversity among different filters of the same layer to alleviate the low-capability of student network. Moreover, we investigate the effect of several prevailing knowledge for face recognition distillation and conclude that the knowledge of feature consistency is more flexible and preserves much more information than others. Experiments on a variety of face recognition benchmarks have revealed the superiority of our method over the state-of-the-arts."