Efficient Dance Form Identification with CNN and SVM: Analyzing Ballet, Hip-Hop, Bhangra, Popping, and Classical Styles

One interesting example of how technology and the arts are interacting is the classification of dance styles by machine learning methods. This research provides a hybrid model to categorize five different dance forms: Ballet, Hip-Hop, Popping, and Classical. The hybrid model combines Convolutional Neural Networks (CNNs) and Support Vector Machines (SVMs). Our model architecture consists of two fully connected layers for classification after four convolutional layers and four max-pooling layers for feature extraction. An extensive collection of dancing photographs was used to assess the model's performance, and it received an overall accuracy of 98%. For all dance forms, the precision, recall, and Fl scores were consistently higher than 95%, and the macro, weighted, and micro averages all lined up at 96.09%. Concerning precision, recall, and Fl-score, classical dance specifically received the best performance metrics, scoring 96.27% overall. These outcomes highlight how well-suited and dependable our CNN-SVM hybrid approach is for managing a variety of dancing styles. The model's potential for useful applications like automatic dance video labeling, personalized suggestions, and interactive training systems is highlighted by its excellent accuracy and consistent performance metrics. This work contributes to the field of dance form categorization by offering a reliable and effective technique that makes use of the advantages of both SVMs and CNNs. More dance forms, bigger datasets, and real-time classification capabilities will all be investigated in future work to improve the model's performance and application.