Method and system for detecting fisheye distorted image

• Main Authors: ZHU HONGTAO, ZHANG ZHENG, XIE SHATING, WEI YUN, LIU JIE, DOU FEI, BAI WENFEI, ZHAO LIYUAN, TIAN QING, ZANG SHUO, NING YAO
• Format: Patent
• Language: Chinese; English
• Published: 16.09.2022
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; PHYSICS

The invention provides a fisheye distortion image detection method and system. According to the invention, a YOLOv5 deep learning model is constructed based on a PyTorch deep learning framework, angle information theta is added in training parameters [x, y, w, h, c] output by the YOLOv5 deep learning model, a rotation frame [x, y, w, h, theta, c] with six-parameter output is obtained, the dimension of a Dataloader module in the YOLOv5 deep learning model is extended to six dimensions to perform six-parameter reading on the rotation frame, and the rotation frame is subjected to six-parameter reading. And obtaining a YOLOv5 rotating frame model suitable for the fisheye image. After the model is trained, distortion in a fisheye distortion image can be overcome through a rotating frame in the internal operation process of the model through a rotating frame algorithm, the detection precision and the detection speed are both considered, and a better detection effect is obtained. 本申请提供一种鱼眼失真图像的检测方法及系统。本申请基于PyTorch深度