Pig Monitoring Algorithm Based on Video Instance Segmentation
ZHANG Changheng
XUE Hui
CHEN Zhong
Abstract：With the vigorous development of artificial intelligence,the intelligent breeding industry has become a new devel-opment prospect.Aiming at the problem of low automation and intelligence of pig farm surveillance video monitoring,a pig monitor-ing algorithm framework based on video instance segmentation is proposed,and it is named Track-YOLACT.The framework is com-posed of two parts:a segmenter and a tracker.First,segmentation is performed based on the classic single-stage instance segmenta-tion network,and then the tracker is used for tracking,so as to realize the segmentation and tracking of all pigs in the video.It pro-vides a practical framework for automated and intelligent monitoring of farms.
Keywords：deep learningvideo instance segmentationintelligent breeding industry
Publication Date：2023-11-20
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：5( 2500-2504 )