Image Recognition Of Diarrhea Lambs Based On Deep Learning

The lamb breeding industry in the newborn lamb acute digestive tract infectious disease,fast onset,high mortality,difficult to detect and treat in time,dysentery and diarrhea is the most obvious feature of lambs.At present,diarrhea lambs are basically dependent on manual,and it is difficult to achieve timely and high coverage.This thesis is based on deep learning and target detection knowledge.It aims to solve this problem through intelligent testing.The main research content and results are as follows:(1)Investigate domestic and foreign scholars on the prevention and control methods and technical methods of diarrhea lambs,determine the target test to identify the diseased lambs,go deep into farmers to collect and mark the 2275 diarrhea lamb image samples.The image sample set is common and complete,which can be used for diarrhea Target detection of lambs.(2)Use the classic Faster-RCNN target detection algorithm,Retinanet algorithm and YOLOV5 algorithm to complete the morbid lamb target detection.(3)Comparing the detection process and results of Faster-RCNN,Retinanet and YOLOV5,using the model evaluation index m AP,YOLOV5 has an average average precision of 95.86%,which is superior to the classic Faster-RCNN algorithm and Retinanet algorithm in terms of precision and recall rate,and the detection speed Also faster.This article uses artificial intelligence target detection technology in the diagnosis of animal diseases in breeding,especially the lamb diarrhea target testing.