期刊
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
卷 33, 期 5, 页码 2181-2194出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/TNNLS.2020.3044078
关键词
Shape; Heating systems; Correlation; Faces; Feature extraction; Face recognition; Active shape model; Boundary-adaptive regression; feature correlations; heatmap regression; heavy occlusions; shape constraints
类别
资金
- National Natural Science Foundation of China [62002233, 62076164, 61802267, 61976145, 61806127, 61732011]
- Shenzhen Municipal Science and Technology Innovation Council [JCYJ20180305124834854, JCYJ20190813100801664]
- Natural Science Foundation of Guangdong Province [2019A1515111121, 2018A030310451, 2018A030310450]
- China Postdoctoral Science Foundation [2020M672802]
This article proposes a multiorder multiconstraint deep network (MMDN) for learning more powerful feature correlations and shape constraints, including an implicit multiorder correlating geometry-aware (IMCG) model and an explicit probability-based boundary-adaptive regression (EPBR) method. Experimental results demonstrate the superior performance of MMDN on challenging benchmark data sets.
Recently, heatmap regression has been widely explored in facial landmark detection and obtained remarkable performance. However, most of the existing heatmap regression-based facial landmark detection methods neglect to explore the high-order feature correlations, which is very important to learn more representative features and enhance shape constraints. Moreover, no explicit global shape constraints have been added to the final predicted landmarks, which leads to a reduction in accuracy. To address these issues, in this article, we propose a multiorder multiconstraint deep network (MMDN) for more powerful feature correlations and shape constraints' learning. Especially, an implicit multiorder correlating geometry-aware (IMCG) model is proposed to introduce the multiorder spatial correlations and multiorder channel correlations for more discriminative representations. Furthermore, an explicit probability-based boundary-adaptive regression (EPBR) method is developed to enhance the global shape constraints and further search the semantically consistent landmarks in the predicted boundary for robust facial landmark detection. It is interesting to show that the proposed MMDN can generate more accurate boundary-adaptive landmark heatmaps and effectively enhance shape constraints to the predicted landmarks for faces with large pose variations and heavy occlusions. Experimental results on challenging benchmark data sets demonstrate the superiority of our MMDN over state-of-the-art facial landmark detection methods.
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