Stair Recognition using CNN and Step Detection using Depth Camera

doi:10.50987/jsod.21.1_78

Traffic lights, stairs and steps are a significant barrier restricting activities for the visually impaired. Training a guide dog requires a certain amount of time and a specialized trainer. Regarding the domestic situation, the spread of guide dogs is still insufficient. Therefore, this study conducts fundamental research to develop walking navigation system for the visually impaired by applying image analysis and machine learning recognition technology. Our previous research demonstrated that our recognition system is useful in detecting pedestrian traffic signals, although we still need to adjust some elements before practical use. In this study, we examined some technologies for recognizing stairs and steps for the visually impaired by organizing existing technologies. This paper aims to provide guidelines for designing low-cost walking navigation for the visually impaired by combining existing small computers premised on widespread use and existing IoT and machine learning technologies. We figure out what technologies and components we should combine to achieve the system. Also, we propose a design guideline for developing a navigation system.

Stair Recognition using CNN and Step Detection using Depth Camera

Bibliographic Information

Abstract

Journal

Details 詳細情報について

Export

Report a problem