オプティカルフローの角度分布に基づいた作業現場における動体検出

抄録

In recent years, there has been a remarkable development in human detection technology. Among them, Haar-like feature is known as one of the excellent methods. However, depending on the usage environment, this method may not work well in some cases. Therefore, in this paper, we describe a method for distinguishing between people and cars at a construction site. The distinction is done by measuring the angle distribution from the motion of a person and a car by Optical Flow, and letting the Neural Network learn the difference in the angle distribution. The conditions of the person were working, moving in a half-sitting posture, and kicking a cart.