Abstract: Bharatanatyam, among India's most ancient classical dance traditions, employs intricate hand mudras and facial expressions as fundamental elements of narrative expression, whose accurate interpretation conventionally demands extensive training under expert supervision. This research proposes an intelligent web-based Bharatanatyam Mudra and Emotion Detection System that automates recognition through advanced deep learning and artificial intelligence techniques applied to input images. The architecture features a custom-trained ConvNeXtV2 convolutional neural network for precise mudra classification, complemented by the Google Gemini AI API for comprehensive facial emotion analysis, delivering instantaneous predictions with confidence metrics and interpretive descriptions via a robust Flask backend integrated with MySQL for secure data persistence. Facilitating user authentication, image upload, real-time inference, and historical result retrieval, the system empowers dance practitioners, educators, and scholars with an accessible platform for systematic analysis. By synergistically fusing classical artistic heritage with contemporary computational intelligence, this framework advances cultural preservation while enabling interactive, technology-enhanced pedagogical and analytical methodologies.

Keywords: Bharatanatyam, Hand Gesture Recognition, Facial Emotion Analysis, ConvNeXtV2, Machine Learning, Artificial Intelligence, Deep Learning, Flask Web Application

Downloads: | DOI: 10.17148/IARJSET.2026.13241