Recognition and Classification of Various Dance Forms by Using SVM Classifier

Abstract

Indian traditional dance involves a tough and difficult dance poses and hand gesture and leg movements, so it is a tedious task to recognize and discover human dance movements from video sequences because dance poses are hidden in their heavy makeup and typical costumes and it make nervures for audience who watching dance performance and also to dance fans. In this paper work we have taken into account this problem and made an improvement by using a new classification model. In this new model we demonstrated three varieties of classical dance forms such as Bharathanatium, oddisi, kathak. In Research lab we experimented various form of dance videos and simultaneously carried an investigation on frames (images) generated from uploaded video by using a python integrated shell environment. The new model used a Shannon entropy algorithm for preprocessing of particular dance image and applied a Hu movements for extract a features and this will be issued as input to SVM classifier and then it predict which dance form it is and at same time it display a pixel value of that image with this new model system performance is high recognition accuracy is 94.4% with SVM classifier.