Scholarlink

Indian classical dance has its roots in Abhinaya, the skill of telling stories through expression where tiny emotions are key to showing the Navarasa (aesthetic states). Yet, the usual way of judging these quick emotional shifts is still based on personal views. This paper presents NaMENet (Navarasa related Micro-Emotion Network), a system that uses deep learning to analyze these small emotions in Indian classical dance without bias. NaMENet combines FAU (Facial Action Unit detection) detection with time-based modeling to group six main emotion types. A two-way recurring setup makes sure emotions flow over time catching quick feeling shifts that match Abhinaya. Tested on a selected Indian classical dance dataset, NaMENet reaches an accuracy of 91.6%, a Precision-Recall AUC of 0.89, and an F1-score of 0.90. This system connects the traditional teacher-student evaluation with AI-powered dance education providing a scalable and unbiased method to analyze tiny emotions in the performing arts.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Facial Action Unit Analysis for Indian Classical Dance Forms: A Micro-emotion Sequencing Framework Approach

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Manav Bakliwal,
  • Dhruv Pandey,
  • Saswata Kumar Dash,
  • S. Kanimozhi

摘要

Indian classical dance has its roots in Abhinaya, the skill of telling stories through expression where tiny emotions are key to showing the Navarasa (aesthetic states). Yet, the usual way of judging these quick emotional shifts is still based on personal views. This paper presents NaMENet (Navarasa related Micro-Emotion Network), a system that uses deep learning to analyze these small emotions in Indian classical dance without bias. NaMENet combines FAU (Facial Action Unit detection) detection with time-based modeling to group six main emotion types. A two-way recurring setup makes sure emotions flow over time catching quick feeling shifts that match Abhinaya. Tested on a selected Indian classical dance dataset, NaMENet reaches an accuracy of 91.6%, a Precision-Recall AUC of 0.89, and an F1-score of 0.90. This system connects the traditional teacher-student evaluation with AI-powered dance education providing a scalable and unbiased method to analyze tiny emotions in the performing arts.