Indian Sign Language Recognition using AlexNet

摘要

Sign language is the language which people with hearing problems used to communicate. The problem will be more complicated if the listener is does not understand the sign language. Here in this project we made use of image based hand gesture recognition techniques for identifying the sign language. The input sequence of images are captured with a digital camera. It is then passed through a sequence of preprocessing steps for the removal of background noise. The image is segmented based on the skin colour and is later on edge segmented to obtain the exact shape of the detected skin pixel. Canny edge detector is used for this purpose. The recognition of hand or body with surroundings is done by comparing the brightness information in HSI colour model. The colour thresholded image is then given to the AlexNet and the gesture signaled is identified in to a particular category and the label is extracted.