Research on Video Processing Based on YOLOv3 Improved Algorithm

YOLO series algorithms are deep learning algorithms widely used in visual inspection and have a significant influence on processing remote sensing images and monitoring images. Compared with pictures, it tests the processing speed of algorithms and the accuracy of judging complex environments in the field of video processing. In this paper, using YOLOv3 and TinyYOLOv3 respectively, we design and improve the YOLOv3 algorithm, which makes the new algorithm increase the judgment accuracy based on relatively enhancing the operation speed, and improve the processing frame rate in the same video processing and solve the problem of recognition error under the occlusion of moving objects.