This demonstration showcases a novel system for micro-expression recognition (MER). MER is crucial for affective computing, enabling the detection of hidden emotions through involuntary facial movements. However, micro-expressions are fleeting and subtle, posing challenges for conventional cameras. Our system addresses this by employing event cameras to capture these rapid, nuanced expressions. The system comprises two key components: the Event-Enhanced Motion Extractor (EEME) amplifies the detection of subtle movements, while the Event-Guided Attention (EGA) focuses on crucial facial regions for micro-expression analysis. This work presents a valuable tool for researchers and practitioners in the field of affective computing, which is associated with an accepted paper of ICME2024 main track.
本文提出了一种基于事件相机的微表情识别系统。针对微表情持续时间短、幅度微弱、难以用传统相机捕捉的问题，系统利用事件相机的高时间分辨率特性，设计了事件增强运动提取器（EEME）以放大细微运动，并引入事件引导注意力（EGA）聚焦关键面部区域，从而提升微表情识别的准确性与鲁棒性。该系统为情感计算领域提供了有效工具。