Recognition Aid for Visually Challenged to Make Out Indoor Environment

Abstract

The applications that are capable of identifying the tangible products and the movement of those product has attained a predominant position in the field of robotics. The complexity in identifying the changes going around is very high on the indoor environments that are too messy. Segregating things and sorting objects in such messy environment becomes even more tedious and challenging for the people with visual disorders. To subdue these issues and enable the blind and the visually challenged to be aware of the changes or the tangible objects they come across in the indoor environment, the proffered method in the paper devise a recognition aid that is empowered with the deep learning neural networks. The usual conventional-CNN is refurbished by upgrading the components to achieve a better accuracy in recognizing. The images based on different scenes of the indoor environment gathered under different circumstance where used as training and the testing dataset for the proffered model and the accuracy and the recognition rate for the training as well as testing was examined.