Artificial Neural Network based Indian Sign Language Recognition using hand crafted features
Purva C. Badhe, Vaishali Kulkarni
Abstract
Sign language is a medium of communication of Divyangjans (Deaf and mute people). Sign language can be used for effective communication only if understood by both the people trying to communicate. When one person is unaware of meaning of the sign gestures an interpreter is required who can translate the gestures into spoken language. This article presents a methodology to recognize Indian Sign Language (SL) gestures and translate them into English. SL Recognition systems can be useful for facilitating the conversation. There are various systems developed by researchers for implementing a SL recognition system. Being in its developing stage, the grammar rules of Indian Sign Language (ISL) are not documented making the recognition process a challenge. This approach employs hand crafted feature extraction technique and uses Artificial Neural Network for classification of the gestures. The accuracy of model achieved is as high as 98% using this methodology.