SOFM and T1FS based measurement of correctness in dance postures for e-learning applications
Sriparna Saha, Rimita Lahiri, Sanchita Purohit Ghosh, Amit Konar · 2016
Dance posture recognition has emerged as one of the most enriched and pragmatic research genre because of its wide applications for facilitating learning of dance by exploiting electronic means only. Ballet dance is one of the most ancient dance forms; moreover the artistic postures and unprecedented elegance of the ballet dance form fascinate the dance enthusiasts a lot. This motivated us to design a system enabling e-learning of ballet dance that allows a user to learn the art all by himself with the help of associative devices and to correct the postures by measuring the extent of correctness, thus not requiring the presence of any instructor to identify the flaws. In this paper, the principles of Type 1 Fuzzy Inference rules have been embedded in the correctness measurement phase of Self Organizing Feature Map. This paper primarily employs Self Organizing Feature Map because it serves the purpose of dance posture recognition and correctness measurement employing Type 1 Fuzzy rules inside a single hybrid framework. This scheme deals with 20 different body joint oriented features covering entire human skeleton and thus provides significantly better results. After analyzing and comparing the experimental findings it can be easily inferred that the designed scheme surpasses the existing methodologies by a noticeable margin.