4DMR 2025
4D Micro-Expression Recognition Workshop at IJCAI 2025

Table of Contents

• The 1st Workshop & Challenge on 4D Micro-Expression Recognition for Mind Reading (4DMR) was held in conjunction with IJCAI 2025 on 29 August 2025, jointly in Montreal, Canada and Guangzhou, China. 4DMR provided a timely platform for advancing 4D facial expression research. It featured the first 4D Micro-Expression Recognition Challenge, enabling a public benchmark for evaluating emerging technical solutions. The workshop covered a broad range of topics, including: (1) Detection and analysis of macro- and micro-expressions for emotion understanding; (2) 4D-based recognition, online analysis, and generative modeling; (3) Psychologically grounded mechanisms of facial behaviors related to cognition; (4) Robust learning in complex in-the-wild scenarios; (5) Multimodal behavioral signals for cognitive state inference; (6) Social and clinical applications such as mental health and education; (7) New datasets for mind-reading research; (8) Realistic expression generation for human–AI interaction. There were 12 papers submitted for peer-review to this workshop. Out of these, 8 papers were accepted for this volume, 3 as regular papers and 5 as short papers. This workshop was organized and supported by the EU Horizon-MSCA Staff Exchange project ACMod, which includes a structured ethical framework and compliance monitoring. All accepted papers involve basic research only; no technology is presented for deployment or operational use. The datasets involved are publicly available and compliant with GDPR, and the work follows the EU Ethical AI guidelines as well as project-level oversight by an independent Ethics Committee. We thank all authors, reviewers, challenge participants, invited speakers, and organizing partners for their contributions. Their efforts have established a strong foundation for future progress in 4D micro-expression analysis and facial behavior understanding.

Session 1: Oral papers

• A Micro-Expression Recognition Method Based on an Uncertainty-Aware Mixing Strategy and Multimodal Fusion 1-9
  Qian Gao, Weijia Feng, Jia Guo, Jiayi An, Xiaofeng Wang, Yuanxu Chen
• DIANet: A Phase-Aware Dual-Stream Network for Micro-Expression Recognition via Dynamic Images 10-19
  Vu Tram Anh Khuong, Luu Tu Nguyen, Thi Bich Phuong Man, Thanh Ha Le, Thi Duyen Ngo
• Inference-Time Gaze Refinement for Micro-Expression Reconition Enhancing Event-Based Eye Tracking with Motion-Aware Post-Processing 20-43
  Nuwan Bandara, Thivya Kandappu, Archan Misra
• Temporal Related Transformer with Depth Information for 4D Micro Expression Recognition 44-51
  Jinsheng Wei, Jialiang Sun, Guanming Lu, Jingjie Yan
• Leveraging Foundation Models and 3D Facial Reconstruction for Micro-Expression Classification 52-58
  Le Cong Thuong, Hai-Chau Nguyen-Le, Tu Nguyen Luu, Thi Duyen Ngo, Thanh Ha Le
• Three-Stream Region-Aware Residual Attention Network for Facial Depression Recognition 59-68
  Muhammad Turyalai Khan, Faisal Shafait, Pradorn Sureephong
• Impact of Trial-wise and Test Data Leakage on EEG-Based Emotion Classification 69-77
  Peihong Lei, Mengyao Wu, Wenjun Yig, Hanlin Mo
• A Conceptual Framework for Emotion-Aware Monitoring of Sundown Syndrome 78-85
  Qianru Xu, Penglan Liu, Chaoxiong Ye