Deep Learning based Detection of Mobility Aids using YOLOv5
A. Siva Kailash, B. Sneha, M Authiselvi, M Dhiviya, R. Karthika, E. Prabhu
Abstract
Mobility impairments and difficulties are a problem faced by a significant portion of the population. The statistics on differently abled people and the mobility aids can benefit a great deal for governmental institutions in areas such as policy-making and infrastructure development. This paper proposes an automated process for the detection and identification of people who use mobility aids. The YOLO (You Only Look Once) algorithm is used to train the dataset. The performance metrics and the feasibility of the algorithm are being discussed. The primary aim of this paper is to help improve the facilities made for differently people in public places by the detection and analysis of the different kinds of mobility aids that people use and how it affects movement in a public place. The model detects and classifies each person in the image as differently abled or not, and if found to be differently abled, it finds out the type of the mobility aid used and the number of differently abled people in each class as well.