Get Started
Home
Topics
Search
Library
Research questionHow can skeleton-based emotion recognition generalize to unseen performers while exposing body-motion evidence under leave-performer-out evaluation?Skeleton motion varies across people, so a classifier can rely on performer-specific cues and fail on unseen performers. Even when predictions improve, it may remain unclear whether they reflect meaningful body-region motion or diffuse temporal patterns.
AI
Computer Vision
Evaluation & Benchmarks
Machine Learning
Research Paper
Statistical Machine Learning
Technology
Latest papersRecent research connected to this question, newest first.Orthogonal Ensembles and Tested Explanations for Performer-Independent Body-Motion Emotion RecognitionThe evidence concerns body-only, 12-class acted-emotion classification from skeleton motion using leave-performer-out evaluation. It reports an 11-model logit-mean ensemble and post hoc part-masking and counterfactual audits, including comparisons between region-level saliency and Laban Movement Analysis attributes; the temporal-saliency result is diffuse, and the audit requires no retraining.research paper · Sep 2, 2026
Related questions
How can real-time emotion recognition in VR handle upper-face cues occluded by head-mounted displays?How can speech and facial cues be combined for emotion recognition when timing and class balance vary?How can conversational speech emotion recognition use cross-speaker context without confusing each speaker’s emotional trajectory?How can video models recognize unseen actions from only a few labeled examples?