New SAC2-Net improves micro-expression recognition with multimodal fusion

Researchers have developed SAC$^2$-Net, a novel network designed to improve micro-expression recognition (MER) by addressing the challenges of subtle facial movements and limited data. The network leverages the complementary nature of optical flow and motion magnification, which often capture different aspects of facial dynamics. SAC$^2$-Net employs Semantic Anchoring Soft Alignment (SASA) to align these modalities using textual prompts derived from Action Units (AUs) as semantic anchors. It then utilizes Complementary-Consensus Fusion (CCF) to refine the fused representations by exchanging unreliable evidence and enforcing a shared spatial focus. Experiments on five benchmarks demonstrate that SAC$^2$-Net achieves state-of-the-art performance in various MER evaluation settings. AI

IMPACT This research could lead to more accurate and nuanced emotion detection systems, with potential applications in human-computer interaction and behavioral analysis.

RANK_REASON The cluster contains a research paper detailing a new model and methodology for a specific AI task. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CV →

• Action Units (AUs)
• Complementary-Consensus Fusion (CCF)
• Micro-expression recognition (MER)
• Motion magnification for endoscopic surgery
• optical flow
• SAC$^2$-Net
• Semantic Anchoring Soft Alignment (SASA)

• paper
• model release

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