Research on the Influence of Dehazing Algorithm on YOLOv3 Target Recognition

Abstract

The YOLOv3 target recognition algorithm has a wide range of applications in various fields. Under the haze conditions, the recognition effect of the YOLOv3 algorithm is affected, and its mAP value decreases significantly. To solve this problem, it can be improved by adding a dehazing algorithm before the recognition algorithm. In this paper, three commonly used dehazing algorithms are studied, and they are combined with YOLOv3 algorithm to perform target recognition experiments on hazing images. Solve their mAP values separately and compare them. The results show that all the three dehazing algorithms can improve the target recognition ability, and the Retinex algorithm works best.