Abstract
This paper describes a visual object recognition method applicable for smart devices to identify crosswalks and pedestrian signals. The device, which is incorporated into a grip of handheld equipment, is capable to support users which have difficulty to recognize crosswalks. In this paper, a novel object detection methodology is proposed, which is a combination of CNN and computational visualization technology. By using the proposed object detection methodology, a smart assistant system for crossing roads is developed. The developed method exhibited highly accuracy recognition of the target objects.
