Preface The 3rd MiGA Workshop & Challenge to explore using body gestures for both hidden and general emotional state analysis (MiGA3 in short) was jointly hosted at the IJCAI 2025 conference, in Guangzhou, China.
As introduced in the 1st and 2nd MiGA workshops, we focus on a specific group of body gestures, called micro-gestures (MGs), used in the psychology research field to interpret inner human feelings. With more and more research attention drawn to Micro-gestures, we continue to organize the third workshop focusing on human behaviour and furthermore, a newly introduced task: how to utilize human behaviors for emotion understanding this year.
The third MiGA workshop and challenge seeks to broaden the research community focused on human behaviour analysis and its applications in emotion understanding by introducing a brand new task in the MiGA series, MG-based emotion understanding. The event aims to foster dialogue among researchers from academia and industry, highlighting key attributes influencing gesture-based emotion recognition and evaluating recent advancements in the field. Similar to the first and second MiGA, we introduce two distinct datasets (SMG and iMiGUE datasets) and corresponding benchmarks (MG classification, online recognition, and a new task, MG-based emotion recognition), with the goal of shaping a new trajectory for the emotion AI community.
Building on the success of its inaugural event, MiGA 2025 was organized as a half-day workshop in Guangzhou, China. The workshop featured one invited talk and addressed topics spanning the theoretical foundations, technological advancements, and practical applications of gestures and micro-gestures in emotion understanding. The MiGA 2025 program, hosted in conjunction with IJCAI 2025, included a distinguished invited speaker: Assoc. Prof. Zheng Lian from the Institute of Automation, Chinese Academy of Sciences, China. Additionally, 11 full papers were presented during the workshop, selected through a rigorous peer-review process. An invited paper to summarize the workshop is included later.
We extend our heartfelt thanks to Assoc. Prof. Zheng Lian for his insightful and thought-provoking talks. We are equally grateful to all the participants for their invaluable contributions, which were instrumental in making MiGA 2024 a remarkable event and a dynamic forum for knowledge exchange within the community. Their engagement sparked vibrant discussions on pivotal and contemporary advancements, highlighting an exceptional program that exemplified cutting-edge work at the intersection of AI and emotion AI. Special thanks are also given to Associate Prof. Xiaobai Li and Dr. Yante Li for assisting with this event. We look forward to the opportunity to host future events of this caliber, continuing to foster innovation and collaboration in this exciting field. The following full papers presenting original research works were accepted, and we divided them into three sessions based on the content of the work.
In Session 1: MG Classification and Online Recognition, five papers are included in this session, reporting their competition-winning schemes for the MG recognition tasks.
In Session 2: MG-based Emotion Recognition, three papers are included in this session, reporting their competition-winning schemes for the MG-based emotion recognition task.
In Session 3: Human Behaviour Analysis for Emotion Understanding, four papers are included in this session, bringing a broader spectrum of the research entry in human behaviour analysis for emotion understanding, as well as a summary paper of the workshop and challenge.
The authors have not employed any Generative AI tools for content generation.
Large-scale emotion understanding for in-the-wild human-computer interaction scenarios (Assoc. Prof. Zheng Lian from the Institute of Automation, Chinese Academy of Sciences, China)
Fang Kang (University of Oulu, Finland) Björn W. Schuller (University of Augsburg, Germany; Imperial College London, London/UK)