Long-term Motivation Tracing

Analyzing persistent emotional states beyond instantaneous expressions

Our research in long-term motivation tracing focuses on understanding and analyzing persistent emotional states and underlying motivations that extend beyond momentary emotional expressions. This approach enables comprehensive emotional understanding by considering temporal dynamics and contextual factors in emotion analysis.

Research Focus: Developing de-identified multimodal emotion recognition and reasoning systems that can trace emotional motivations across extended time sequences while preserving privacy through identity-free analysis.

Key Contributions

  • DEEMO Framework: De-identity multimodal emotion recognition and reasoning for privacy-preserving emotional analysis
  • EALD-MLLM System: Emotion analysis in long-sequential and de-identity videos with multimodal large language models
  • Identity-free AI Emotion Intelligence: Comprehensive micro-gesture understanding for tracing emotional motivations
  • Sequence Reasoning: Leveraging temporal patterns for action anticipation and emotional state prediction
DEEMO: De-identity Multimodal Emotion Recognition and Reasoning
D. Li, B. Xing, X. Liu*, B. Xia, B. Wen, and H. Kälviäinen
ACM International Conference on Multimedia (ACMMM), 2025
EALD-MLLM: Emotion Analysis in Long-sequential and De-identity videos with Multi-modal Large Language Model
D. Li, X. Liu, B. Xing, B. Xia, Y. Zong, B. Wen, H. Kälviäinen
Preprint
Identity-free Artificial Emotional Intelligence via Micro-Gesture Understanding
R. Gao, X. Liu*, B. Xing, Z. Yu, B. W. Schuller, and H. Kälviäinen
IEEE Transactions on Affective Computing, 2025
From Recognition to Prediction: Leveraging Sequence Reasoning for Action Anticipation
X. Liu, C. Hao, Z. Yu, H. Yue, and J. Yang
ACM Transactions on Multimedia Computing, Communications, and Applications, 2024
« Back to Research Framework