Study of Filtering the Weather Adverse Effects to Object Detection

The work is devoted to the study of the problem of detecting objects of various sizes on the example of an open dataset using the Yolo v5 [1] neural network model. The main attention is paid to the study of the effect of pre-filtering of images on the results of object detection and the development of a methodology for assessing such an effect. In addition, the paper etestuates the effect of filtering distortions from rain and snow on the results of object detection according to the proposed method. The results obtained can be useful for improving the accuracy of detecting objects in images with various types of distortion and are applied in various fields, such as automatic car driving, transport monitoring, etc.