Research On Pig Detection Tracking And Pig Ear Recognition Algorithm Based On Deep Learning

With the application of intelligent management and camera equipment in the entire pig industry chain,management on an individual basis has gradually become the mainstream,which makes real-time pig individual identification and tracking in video streams an urgent need.At the same time,the pig’s ear is often used as an important inspection site for pig disease identification,and is also the main wearing site for pig markers such as quarantine ear tags,which makes high-quality pig ear images have broad application prospects,which makes the detection of pig ear targets more effective.important meaning.In order to solve the above problems,this paper explores the algorithm based on deep learning in three aspects: pig target detection,pig target tracking,and pig ear target detection.This paper mainly does the following work:(1)A simple pig data collection system was built.(2)Research on the problem of multi-pig target detection and tracking,design and produce a data set according to the use requirements and use scenarios of this specific problem,and train a pig target detection network based on the YOLOv5 s network on this basis,and and The performance comparison and comprehensive analysis of pig recognition models based on other networks are carried out,and then the field video is used to simulate field detection experiments and analyze.(3)For the pig target tracking problem,through the analysis and judgment of specific scenes,for the scene characteristics such as the difficulty of matching the front and rear frames caused by the mutual occlusion of pigs,the traditional Deep SORT algorithm has been improved in a targeted manner,and then based on the different motion conditions of the two segments The video production pig tracking test set was used to test and analyze the performance of the algorithm before and after the improvement.(4)For the detection of pig ear parts,the robustness of the algorithm is improved by comprehensively constructing a data set and enhancing the data set.At the same time,by making targeted improvements to the SSD network,the detection efficiency of pig ear parts is improved.The performance of the network before and after the improvement is compared and analyzed.The research results show that the pig target detection algorithm based on the YOLOv5 s network has excellent real-time performance and recognition performance;compared with the original algorithm,the improved Deep SORT algorithm can achieve better real-time performance in pig target tracking problems.Tracking performance: The pig ear part detection algorithm constructed by the improved SSD network has good reliability,accuracy and real-time performance.