AI 中文总结
针对基于身体运动的骨架情感识别难题，在MMAC ACII 2026挑战赛DIEM - A任务中，提出多分支框架，含基于6D旋转、部分感知动力学多流及元数据条件弱标签分布学习分支，经独立训练与概率级融合，提升了准确率和宏F1，揭示重要运动线索。
AI 中文摘要
基于身体运动的骨架情感识别具有挑战性，因为情感表达常由微妙的动态和关系运动线索表征，硬标签可能无法完全捕捉相关情感类别间的模糊性。针对MMAC ACII 2026挑战赛中的DIEM - A任务，我们提出一个多分支基于骨架的情感识别框架，它结合了基于6D旋转的分支、部分感知动力学多流分支和元数据条件弱标签分布学习分支。各分支独立训练，推理时通过概率级融合。在10折留一表演者交叉验证中，该框架将基于旋转的基线的准确率从0.271提高到0.366，宏F1从0.252提高到0.353。可解释性消融表明速度和骨骼流以及手臂和腿部区域为识别情感身体运动提供重要线索。
英文摘要
Skeleton-based emotion recognition from body motion remains challenging because emotional expressions are often characterized by subtle dynamic and relational motion cues, and hard labels may not fully capture ambiguity among related emotion categories. For the DIEM-A task in the MMAC ACII 2026 Challenge, we propose a multi-branch skeleton-based emotion recognition framework that combines a 6D rotation-based branch, a part-aware kinetic multi-stream branch, and a metadata-conditioned weak label distribution learning (LDL) branch. The branches are trained independently and fused by a probability-level ensemble at inference time. In 10-fold leave-performer-out cross-validation, the proposed framework improves Accuracy from 0.271 to 0.366 and Macro-F1 from 0.252 to 0.353 over the rotation-based baseline. Explainability ablations show that velocity and bone streams, as well as arm and leg regions, provide important cues for recognizing emotional body motion.
CommentsAccepted to ACII2026 workshop
