Indoor place recognition for the blind

Navigating inside building is a challenging task for the blind which often makes them dependent on others foe guidance. This study explores place recognition accessible via hand held devices to guide the visually impaired inside unfamiliar buildings. In computer vision, place recognition addresses the problem of identifying places based on appearance. The majority of this study focuses on implementing, evaluating, and enhancing place recognition algorithms for this application domain. The outcome of this study is a camera-based indoor blind-navigation model that can autonomously recognize previously mapped indoor locations. In the mapping or calibration phase, images of an indoor setting are collected, labeled, and mapped to specific locations in the indoor environment. The effectiveness of the system is evaluated by measuring the recall rate of the place recognition algorithm.

000-099