National Engineering Laboratory of Deep Learning Technology and Application

Facial age estimation has attracted considerable attention owing to its great potential in applications. However, it still falls short of reliable age estimation due to the lack of sufficient training data with accurate age labels. Using conventional semi-supervised methods to exploit unlabeled data appears to be a good solution, but it does not yield sufficient performance gains while significantly increasing training time. Therefore, to tackle these problems, we present a Divergence-driven Consistency Training (DCT) method for enhancing both efficiency and performance in this paper. Following the idea of pseudo-labeling and consistency regularization, we assign pseudo labels predicted by the teacher model to unlabeled samples and then train the student model on labeled and unlabeled samples based on consistency regularization. Based on this, we propose two main promotions. The first is the Efficient Sample Selection (ESS) strategy, which is based on the Divergence Score to select effective samples from massive unlabeled images to reduce the training time and improve efficiency. The second is Identity Consistency (IC) regularization as the additional loss function, which introduces a high dependency of aging traits on a person. Moreover, we propose Local Prediction (LP), which is a plug-and-play component, to capture local semantics. Extensive experiments on multiple age benchmark datasets, including CACD, Morph II, MIVIA, and Chalearn LAP 2015, indicate DCT outperforms the state-of-the-art approaches significantly.

Current semi-supervised semantic segmentation methods mainly focus on designing pixel-level consistency and contrastive regularization. However, pixel-level regularization is sensitive to noise from pixels with incorrect predictions, and pixel-level contrastive regularization has a large memory and computational cost. To address the issues, we propose a novel region-level contrastive and consistency learning framework (RC^2L) for semi-supervised semantic segmentation. Specifically, we first propose a Region Mask Contrastive (RMC) loss and a Region Feature Contrastive (RFC) loss to accomplish region-level contrastive property. Furthermore, Region Class Consistency (RCC) loss and Semantic Mask Consistency (SMC) loss are proposed for achieving region-level consistency. Based on the proposed region-level contrastive and consistency regularization, we develop a region-level contrastive and consistency learning framework (RC^2L) for semi-supervised semantic segmentation, and evaluate our RC^2L on two challenging benchmarks (PASCAL VOC 2012 and Cityscapes), outperforming the state-of-the-art.

Spectral clustering (SC) is an unsupervised learning method that has been widely used in industrial product analysis. Compressive spectral clustering (CSC) effectively accelerates clustering by leveraging graph filter and random sampling techniques. However, CSC suffers from two major problems. First, the direct use of the dichotomy and eigencount techniques for estimating Laplacian matrix's k-th eigenvalue is expensive. Second, the computation of interpolation is time-consuming because it requires to repeat matrix-vector product for every cluster in each iteration. To address these problems, we propose a new method called fast compressive spectral clustering (FCSC). Our method addresses the first problem by assuming that the eigenvalues approximately satisfy local uniform distribution, and addresses the second problem by recalculating the pairwise similarity between nodes with low-dimensional representation to reconstruct denoised laplacian matrix. The time complexity of reconstruction is linear with the number of non-zeros in Laplacian matrix. As experimentally demonstrated on both artificial and real-world datasets, our method significantly reduces the computation time while preserving high clustering accuracy comparable to previous designs, demonstrating the effectiveness of FCSC.

Existing deep learning-based approaches for monocular 3D object detection in autonomous driving often model the object as a rotated 3D cuboid while the object’s geometric shape has been ignored. In this work, we propose an approach for incorporating the shape-aware 2D/3D constraints into the 3D detection framework. Specifically, we employ the deep neural network to learn distinguished 2D keypoints in the 2D image domain and regress their corresponding 3D coordinates in the local 3D object coordinate first. Then the 2D/3D geometric constraints are built by these correspondences for each object to boost the detection performance. For generating the ground truth of 2D/3D keypoints, an automatic model-fitting approach has been proposed by fitting the deformed 3D object model and the object mask in the 2D image. The proposed framework has been verified on the public KITTI dataset and the experimental results demonstrate that by using additional geometrical constraints the detection performance has been significantly improved as compared to the baseline method. More importantly, the proposed framework achieves state-of-the-art performance with real time. Data and code will be available at https://github.com/zongdai/AutoShape

Face presentation attack detection, also termed Face Anti-Spoofing (FAS) [item 1), 2) in the Appendix), is a hot and challenging research topic that has received much attention from the computer vision and pattern recognition communities in the past. Owing to the development of deep learning and big data, recent advances in this and related fields has increased considerably. However, there are still several challenging tasks that deserve attention from the community, for instance robust techniques to unknown spoofing attacks, cross-domain generalization, and multi-modal fusion in images and video sequences. We edited this special issue with the goal of compiling the latest progress in the field and identifying promising research opportunities on FAS.

Abstract Face anti‐spoofing is critical to prevent face recognition systems from a security breach. The biometrics community has achieved impressive progress recently due to the excellent performance of deep neural networks and the availability of large datasets. Although ethnic bias has been verified to severely affect the performance of face recognition systems, it still remains an open research problem in face anti‐spoofing. Recently, a multi‐ethnic face anti‐spoofing dataset, CASIA‐SURF cross‐ethnicity face anti‐spoofing (CeFA), has been released with the goal of measuring the ethnic bias. It is the largest up to date CeFA dataset covering three ethnicities, three modalities, 1607 subjects, 2D plus 3D attack types and the first dataset including explicit ethnic labels among the recently released datasets for face anti‐spoofing. We organized the Chalearn Face Anti‐spoofing Attack Detection Challenge which consists of single‐modal (e.g. RGB) and multi‐modal (e.g. RGB, Depth, infrared) tracks around this novel resource to boost research aiming to alleviate the ethnic bias. Both tracks have attracted 340 teams in the development stage, and finally, 11 and eight teams have submitted their codes in the single‐modal and multi‐modal face anti‐spoofing recognition challenges, respectively. All of the results were verified and re‐ran by the organizing team, and the results were used for the final ranking. This study presents an overview of the challenge, including its design, evaluation protocol and a summary of results. We analyse the top‐ranked solutions and draw conclusions derived from the competition. Besides, we outline future work directions.

Given a set of locations in a city, on which ones should we place ads on so as to reach as many people as possible within a limited budget? Past research has addressed this question under the assumption that dense trajectory data are available to determine the reach of each ad. However, the data that are available in most industrial settings do not consist of dense, long-range trajectories; instead, they consist of statistics on people's short-range point-to-point movements. In this paper, we address the natural problem that arises such data: given a distribution of population and point-to-point movement statistics over a network, find a set of locations within a budget that achieves maximum expected reach. We call this problem geodemographic influence maximization (GIM). We show that the problem is NP-hard, but its objective function is monotone and submodular, thus admits a greedy algorithm with a 1 over 2 (1-1 over e) approximation ratio. Still, this algorithm is inapplicable on large-scale data for high-frequency digital signage ads. We develop an efficient deterministic algorithm, Lazy-Sower, exploiting a novel, tight double-bounding scheme of marginal influence gain as well as the locality proprieties of the problem; a learning-based variant, NN-Sower, utilizes randomization and deep learning to further improve efficiency, with a slight loss of quality. Our exhaustive experimental study on two real-world urban datasets demonstrates the efficacy and efficiency of our solutions compared to baselines.

In Autonomous Driving (AD), detection and tracking of obstacles on the roads is a critical task. Deep-learning based methods using annotated LiDAR data have been the most widely adopted approach for this. Unfortunately, annotating 3D point cloud is a very challenging, time- and money-consuming task. In this paper, we propose a novel LiDAR simulator that augments real point cloud with synthetic obstacles (e.g., cars, pedestrians, and other movable objects). Unlike previous simulators that entirely rely on CG models and game engines, our augmented simulator bypasses the requirement to create high-fidelity background CAD models. Instead, we can simply deploy a vehicle with a LiDAR scanner to sweep the street of interests to obtain the background point cloud, based on which annotated point cloud can be automatically generated. This unique "scan-and-simulate" capability makes our approach scalable and practical, ready for large-scale industrial applications. In this paper, we describe our simulator in detail, in particular the placement of obstacles that is critical for performance enhancement. We show that detectors with our simulated LiDAR point cloud alone can perform comparably (within two percentage points) with these trained with real data. Mixing real and simulated data can achieve over 95% accuracy.