Benchmarking Dynamic Affective Reasoning: A Viewer-Centric Video Emotion Dataset
2026-07-11 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors point out that emotions in videos aren't just isolated feelings but change as events happen one after another. They created a big new dataset called DAR that shows how viewers' emotions evolve across 15,087 videos with detailed labels for 27 different emotions. They set up three tasks for computers to learn: spotting emotional moments, identifying specific emotions, and figuring out why emotions change. To tackle these tasks, the authors introduced a method named DAR-R1, which improves emotion recognition and reasoning better than earlier models.
video emotion analysisaffective reasoningemotion classificationdynamic emotionsbenchmark datasetevent segmentationcausal reasoningmachine learningmulti-modal large language models (MLLMs)
Authors
Zhiyan Zhang, Peipei Song, Jinpeng Hu, Jingyang Jia, Xun Yang, Xiaojun Chang
Abstract
Video emotion analysis is typically framed as a static classification problem, treating each clip as an independent labeled unit. However, such a formulation overlooks a key psychological fact: emotions change as a result of cumulative reactions to consecutive causal events. To bridge this gap, we introduce Dynamic Affective Reasoning, the first large-scale benchmark for viewer-centric affect transitions and causal reasoning over consecutive video events. DAR contains 15,087 videos and 36,908 event-aligned affective segments annotated with 27 emotion categories. Unlike existing video-based emotion datasets, DAR presents a new viewer-centric perspective on fine-grained emotional expressions and transitions, and provides dense, temporally grounded, and causally explicit reasoning chains. Based on DAR, we formally define three challenging tasks: affective segmentation, fine-grained emotion classification, and affective reasoning. Complementing this benchmark, we propose DAR-R1, a two-stage framework that combines supervised fine-tuning with Group Relative Policy Optimization. Experiments across 10+ MLLMs show that DAR-R1 sets a new state-of-the-art for dynamic affective reasoning, in terms of both emotional localization and affective reasoning. Project page: https://github.com/Zhang-Zhiyan/DAR.