Object Detection Using Deep Learning for Visually Impaired People in Indoor Environment

In order to acquaint with the surroundings, spatial perception is essential for a human being. As vision plays a crucial part for locomotion, the visually impaired people find it particularly difficult to move in the surroundings. Obstacle detection and alerting can aid them in the movement. This paper focuses on the development of an electronic white cane to assist the visually impaired people in the indoor environment. The proposed electronic white cane consists of an obstacle detection system using deep learning model. A deep learning-based image classification algorithm is applied on real-time captured image to detect obstacle on the path. The accuracy of the system is reinforced by interfacing ultrasonic sensor. Additionally, infrared sensor is employed to detect small objects near feet. Buzzer and earphones are utilized to alert the visually impaired person. The experimental results exhibit the efficacy of the proposed system.