基于热力图与批次相似性注意力约束的面部表情识别方法
By adopting a facial expression recognition method based on heatmaps and batch similarity attention constraints, the problems of recognition instability and data imbalance under complex conditions in existing technologies are solved, and higher recognition accuracy and robustness are achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SOUTH CHINA UNIV OF TECH
- Filing Date
- 2025-07-23
- Publication Date
- 2026-07-17
AI Technical Summary
Existing facial expression recognition technologies exhibit unstable performance under complex conditions (such as weak expression intensity, complex lighting conditions, occlusion, or pose changes), and the datasets suffer from class imbalance, making it difficult to meet the needs of practical applications.
A facial expression recognition method based on heatmaps and batch similarity attention constraints is adopted. The training model is optimized by fusing features from shallow and deep neural networks and combining consistency loss and intra-batch sample feature similarity attention classification loss.
It improves the model's recognition accuracy and robustness under complex conditions, mitigates the negative impact of data label imbalance, and enhances the ability to recognize tail categories.
Smart Images
Figure CN120877350B_ABST