Image To Text Description Platform For Yakshagana Images By Detecting Crown Using CNN

This research called "Yakshakatha – Image to Text Platform for Yakshagana Images." It employs deep learning to classify crowns in Yakshagana Kirita using CNN classification. Yakshagana is a traditional South Indian art form, well known for its intricate costumes and headgear, posing specific challenges in identifying and classifying crowns. The system aims to classify various Kirita styles and provide detailed descriptions of the respective Yakshagana characters, or Vesha, offering valuable cultural and symbolic insights. The model was trained on a self-labelled dataset of Yakshagana images and used convolutional and pooling layers to capture fine details in crown designs. It achieved an impressive accuracy of 99%, ensuring precise identification and classification. Real-time predictions for the input images are made, which have predetermined mappings and description content for recognized characters. The system employs Google Gemini AI, in order to enrich the text description of the image by integrating cultural background. The responsive interface is provided for users on the image upload, real-time result, and enriched visual feedback using annotated output. This project combines modern image recognition with rich cultural content to preserve and promote the traditional heritage of Yakshagana. It will help the users to understand and appreciate the vibrancy of this art form.