Leveraging face-prior knowledge for general face representation learning

doi:10.1016/j.patcog.2025.111784

	Leveraging face-prior knowledge for general face representation learning
	Sun, Haomiao ; He, Mingjie ; Shan, Shiguang 1; Han, Hu
	2025-12-01
发表期刊	PATTERN RECOGNITION
ISSN	0031-3203
卷号	168 页码:8
摘要	Face analysis, such as face attribute recognition and expression classification, has achieved remarkable progress via supervised deep learning methods. Although effective, supervised learning relies on a large amount of annotated data, which is labor-intensive. Besides, traditional models designed for individual tasks cannot generate a general facial representation that can be applied across multiple tasks. To address these issues, we present a novel general face representation learning method that empowers diverse face analysis tasks by incorporating auxiliary tasks with face-prior knowledge. Our approach employs three auxiliary tasks at pixel, patch, and component levels to exploit information from face images, enabling the model to learn multi-level face representations beneficial to diverse downstream tasks. We evaluate the efficacy of our approach on various downstream face analysis tasks, including facial action unit detection, expression classification, age estimation, and face attribute recognition. Specifically, our method surpasses the state-of-the-art by 3.5% for action unit detection and 1.53% for expression classification. On the AgeDB dataset, our method reduces the mean error by 0.64. Furthermore, in face attribute classification tasks, our model also achieves state-of-the-art performance. Experimental results also show that our proposed method outperforms previous methods under both full-shot and few-shot protocols.
关键词	Face analysis General face representation learning Multi-level face-prior knowledge Auxiliary tasks
DOI	10.1016/j.patcog.2025.111784
收录类别	SCI
语种	英语
资助项目	National Natural Science Foundation of China[U2336213] ; National Natural Science Foundation of China[62306298]
WOS研究方向	Computer Science ; Engineering
WOS类目	Computer Science, Artificial Intelligence ; Engineering, Electrical & Electronic
WOS记录号	WOS:001502403800001
出版者	ELSEVIER SCI LTD
引用统计
文献类型	期刊论文
条目标识符	http://119.78.100.204/handle/2XEOYT63/42330
专题	中国科学院计算技术研究所期刊论文_英文
通讯作者	Shan, Shiguang
作者单位	1.Chinese Acad Sci, Inst Comp Technol, State Key Lab AI Safety, Beijing 100190, Peoples R China 2.Univ Chinese Acad Sci, Beijing 100049, Peoples R China
推荐引用方式 GB/T 7714	Sun, Haomiao,He, Mingjie,Shan, Shiguang,et al. Leveraging face-prior knowledge for general face representation learning[J]. PATTERN RECOGNITION,2025,168:8.
APA	Sun, Haomiao,He, Mingjie,Shan, Shiguang,&Han, Hu.(2025).Leveraging face-prior knowledge for general face representation learning.PATTERN RECOGNITION,168,8.
MLA	Sun, Haomiao,et al."Leveraging face-prior knowledge for general face representation learning".PATTERN RECOGNITION 168(2025):8.