期刊
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
卷 43, 期 11, 页码 4008-4020出版社
IEEE COMPUTER SOC
DOI: 10.1109/TPAMI.2020.2997456
关键词
Face; Detectors; Face detection; Feature extraction; Task analysis; Proposals; Neural networks; Face detection; refinement network; high performance
资金
- National Key Research and Development Plan [2019YFC2003901]
- Chinese National Natural Science Foundation [61872367, 61876178, 61806196, 61976229]
The paper introduces a single-shot refinement face detector named RefineFace, which achieves high-performance face detection through five modules. Experimental results demonstrate that the method achieves state-of-the-art results on multiple datasets.
Face detection has achieved significant progress in recent years. However, high performance face detection still remains a very challenging problem, especially when there exists many tiny faces. In this paper, we present a single-shot refinement face detector namely RefineFace to achieve high performance. Specifically, it consists of five modules: selective two-step regression (STR), selective two-step classification (STC), scale-aware margin loss (SML), feature supervision module (FSM) and receptive field enhancement (RFE). To enhance the regression ability for high location accuracy, STR coarsely adjusts locations and sizes of anchors from high level detection layers to provide better initialization for subsequent regressor. To improve the classification ability for high recall efficiency, STC first filters out most simple negatives from low level detection layers to reduce search space for subsequent classifier, then SML is applied to better distinguish faces from background at various scales and FSM is introduced to let the backbone learn more discriminative features for classification. Besides, RFE is presented to provide more diverse receptive field to better capture faces in some extreme poses. Extensive experiments conducted on WIDER FACE, AFW, PASCAL Face, FDDB, MAFA demonstrate that our method achieves state-of-the-art results and runs at 37.3 FPS with ResNet-18 for VGA-resolution images.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据