Pig checking method based on visual inspection

• Main Authors: GUO LIANGZHI, WANG XIAOBING, ZHU FEI
• Format: Patent
• Language: Chinese; English
• Published: 12.07.2024
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; IMAGE DATA PROCESSING OR GENERATION, IN GENERAL; PHYSICS

The invention discloses a pig checking method based on visual detection, and belongs to the technical field of pig checking. The invention discloses a pig checking method based on visual detection. Comprising the following steps: acquiring a pig video stream; acquiring a video frame; preprocessing the image; carrying out self-attention and target segmentation; target tracking; and checking the pigs. The problems that in the prior art, pigs are recognized only through images, recognition errors are caused due to the fact that the sizes and fur varieties and colors of various pigs are inconsistent, and the checking accuracy is affected due to the fact that illumination is too strong or dark in the field environment are solved, the matching quality is improved, the matching time is saved, and the checking accuracy is improved. According to the method, numerical parameters in a Kalman filtering algorithm can be automatically adjusted, so that covariance matrixes of measurement noise and prediction noise can be ad