Real-Time Detection of Pedestrian Traffic Lights for Visually-Impaired People

摘要: Nowadays there are more than 250 million visually-impaired people worldwide and mobility autonomy in outdoor environments is perhaps the greatest challenge they have to face. More specifically, crossing street with no human aid an open problem, since majority of pedestrian traffic lights underdeveloped countries do not provide sound aids. There very few studies addressing detection based on images acquired by mobile devices, best our knowledge a clear gap literature regarding use recent state-of-the-art computer vision approaches such as deep neural networks for problem. In this paper, we investigate current localization/detection classification networks, present solution together their state helping cross streets devices. For such, novel public dataset 4,399 labeled detailed comparison among methods localization/detection. We show empirical evidence feasibility embedding approach devices so it can be used visual impairment.