基于多模态大模型的游客情绪实时感知与干预方法
By integrating tourists' facial, gait, voice, and body surface data into a multimodal large model, a cross-modal attention mechanism and emotion trend prediction model are constructed. This solves the problem of insufficient tourist emotion perception and intervention in smart tourism systems, realizes real-time emotion recognition and dynamic intervention, and reduces safety risks.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- HANGZHOU KANYUANFANG TECHNOLOGY CO LTD
- Filing Date
- 2026-03-25
- Publication Date
- 2026-07-17
AI Technical Summary
The existing smart tourism management system lacks a fusion understanding of tourists' multimodal data, resulting in insufficient real-time perception and intervention of emotions. In particular, it cannot promptly identify and process subcritical emotional fluctuations in high-density crowd scenarios, posing a safety hazard.
A multimodal large model is adopted to construct a cross-modal attention mechanism and emotion fusion representation vector by collecting facial micro-expression, gait posture, speech and body surface thermal data. Combined with neural frequent differential equations, emotion trend prediction is performed, and multi-level intervention strategies are implemented to form a closed-loop adaptive regulation mechanism.
It achieves high-precision real-time identification and prediction of tourists' emotions, reduces safety risks, has good generalization ability and scalability, and can dynamically intervene and optimize strategies before emotions get out of control.
Smart Images
Figure CN121921852B_ABST