Feature agglomeration networks for single stage face detection

Abstract

Recent years have witnessed promising results of face detection using deep learning. Despite making remarkable progresses, face detection in the wild remains an open research challenge especially when detecting faces at vastly different scales and characteristics. In this paper, we propose a novel simple yet effective framework of "Feature Agglomeration Networks" (FANet) to build a new single stage face detector, which not only achieves state-of-the-art performance but also runs efficiently. As inspired by Feature Pyramid Networks (FPN) [11], the key idea of our framework is to exploit inherent multi-scale features of a single convolutional neural network by aggregating higher-level semantic feature maps of different scales as contextual cues to augment lower-level feature maps via a hierarchical agglomeration manner at marginal extra computation cost. We further propose a Hierarchical Loss to effectively train the FANet model. We evaluate the proposed FANet detector on several public face detection benchmarks, including PASCAL face, FDDB and WIDER FACE datasets and achieved state-of-the-art results. Our detector can run in real time for VGA-resolution images on GPU.

DSFD: Dual Shot Face Detector

“…Besides, Wang et al [32] design RepLoss for pedestrian detection, which improves performance in occlusion scenarios. FANet [37] create a hierarchical feature pyramid and presents hierarchical loss for their architecture. However, the anchors used in FANet are kept the same size in different stages.…”

“…To further address the class imbalance problem, Lin et al [18] propose Focal Loss to focus training on a sparse set of hard examples. To use all original and enhanced features, Zhang et al propose Hierarchical Loss to effectively learn the network [37]. However, the above loss functions do not consider progressive learning ability of feature maps in both of different levels and shots.…”

“…First, we introduce a Feature Enhance Module (FEM) to enhance the discriminability and robustness of the features, which combines the advantages of the FPN in PyramidBox and Receptive Field Block (RFB) in RFBNet [19]. Second, motivated by the hierarchical loss [37] and pyramid anchor [27] in PyramidBox, we design Progressive Anchor Loss (PAL) that uses progressive anchor sizes for not only different levels, but also different shots. Specifically, we assign smaller anchor sizes in the first shot, and use larger sizes in the second shot.…”

Assessment and Estimation of Face Detection Performance Based on Deep Learning for Forensic Applications

“…Recent advances in deep learning methods have contributed to significant performance improvements in a wide range of computer vision applications. They have been particularly successful for face detection problems where modern deep CNN models show a significant accuracy improvement in comparison to traditional approaches based on hand-crafted features [22][23][24][25][26][27][28][29][39][40][41][42][43][44][45]. Consequently, these deep learning methods have become the state-of-the-art for face detection.…”

“…Nowadays, deep learning approaches are widely applied for face detection as they enable the system to automatically learn representations from raw input images using a Convolutional Neural Network (CNN), achieving a high accuracy under very challenging detection conditions [22][23][24][25][26][27][28][29]. Most of the deep-learning-based face detectors are, however, computationally demanding and may not be suitable for applications that analyze large amounts of data and require real-time performance, such as the CSEM detection systems in forensic tools.…”