Image Processing-Based Recognition of Sattriya Dance Hand Mudras using a Hybrid CNN-SVM Model for Pattern Recognition in Signal Processing Applications

Classical dances like Sattriya dance have much cultural and artistic importance, and all the gestures carry a specific meaning or meanings in their forms of hand mudra. There is, however, an exception to this rule in terms of automatic recognition of these hand mudras owing to the different positions of the hands, lighting, and background. In this paper, a hybrid model of Convolutional Neural Networks (CNN) and Support Vector Machines (SVM) has been introduced for the recognition of a single-hand Sattriya dance mudras. The CNN model has the capability of inducting the high-level features in pictures of hand gestures and classifying them through the use of an SVM classifier. A complete dataset of 13 different hand mudras with variations added to their original forms to increase variations was used to train and test the model. Most classes showed high specificity with a precision score, recall score, and F1-score of the model of almost 1.0, meaning the power of the model to classify complicated mudra dance. The offered model shows how AI can be used to automate the identification of traditional dance movements, which can provide perspectives on digitization and preservation of the cultural art.