Indoor Environment Assistance Navigation System Using Deep Convolutional Neural Networks

Abstract

This chapter deals with the development of a smart and intelligent deep learning-based indoor landmark identification system. This work is proposed especially to improve the autonomy and the life quality for blind and visually impaired persons (VIP). The proposed indoor object recognition system highly contributes to build an indoor smart environment for this category of persons and facilitate for them more in life. Indoor object classification presents a key technology for indoor navigation assistance systems. Recognizing indoor objects widely helps VIP in their indoor navigation in order to fully participate in the daily life. A deep convolutional neural network (DCNN) is presented to perform indoor object classification. Indoor object detection from real-world images and videos is a highly recommended area of research in the field of artificial intelligence. In this work, we propose a robust technique for indoor object recognition based on deep CNN model. Experimental results performed on natural images with natural illumination shows that our approach achieves good performances with an accuracy of 99.9% for indoor object classification.