A Novel method for visually impaired using object recognition

Abstract

To perform daily activities is a very difficult task for visually impaired people, since vision plays a central role in almost every activity of ours. With the advancement in computer vision and computing technologies we can afford to develop a system for visually impaired people, which can give audio feedback of surrounding objects and context. In this paper a novel assistive method for visually impaired people using object and color recognition is proposed, which is very useful for their safety, quality life and freedom from the other person all the time. The system is containing two modules (i) object recognition, (ii) color recognition. Object recognition module recognizes objects like door, chair, stairs, mobile phone etc. and generate an audio feedback to the user, we are using convolution neural network prior to recursive neural network which is followed by softmax classifier. Second module of this system is color recognition which tell them about object color in front of the camera, this module is useful in recognizing clothing colors, fruits color, police(from uniform color) etc. We are using HSI color space with observed thresholding for color recognition. System modules and operation can be selected with an on demand push button panel. The algorithm is evaluated on an online dataset as well as on our dataset.