New MEDN Network Decouples Motion and Emotion for Micro-Expression Recognition

Researchers have developed a new Motion-Emotion Feature Decoupling Network (MEDN) to improve micro-expression recognition. This network addresses the challenge that micro-expressions can have similar facial action units but convey opposite emotions. MEDN uses a dual-branch framework to separately extract motion and emotion features, with a Sparse Emotion Vision Transformer (SEVit) for implicit emotion modeling and a Collaborative Fusion Module (CoFM) to adaptively merge these disentangled features. Experiments on benchmark datasets show that MEDN achieves superior recognition performance and generalization. AI

IMPACT This research offers a novel approach to micro-expression recognition by disentangling motion and emotion features, potentially improving accuracy in AI systems that analyze subtle facial cues.

RANK_REASON The cluster contains a research paper detailing a new network architecture for a specific computer vision task. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CV →

• Action Units
• Chenxing Hu
• Collaborative Fusion Module
• Motion-Emotion Feature Decoupling Network
• Sparse Emotion Vision Transformer

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →