基于关键点检测和多目标跟踪的猪只体尺估计
Estimation of pig body measurements based on keypoint detection and multi-object tracking
姚裔芃 1徐晨 1陈鸿基 2刘勇 2徐顺来2
作者信息
- 1. 西南大学计算机与信息科学学院/软件学院,重庆 400700;重庆市畜牧科学院/国家生猪技术创新中心,重庆 402460
- 2. 重庆市畜牧科学院/国家生猪技术创新中心,重庆 402460
- 折叠
摘要
[目的]减少猪场人工测量猪只体尺的工作量,提高测量精度和工作效率.[方法]本研究提出基于关键点检测和多目标跟踪的猪只体尺自动估计方法,该方法使用Yolov8-Pose模型识别各猪只关键点和目标检测框,利用ByteTrack算法对猪群实时跟踪,引入感兴趣区域规避图像畸变,提高识别速度,同时设计姿态和异常检测过滤算法减少因运动模糊、姿态不正等因素造成的误差.[结果]5 个猪栏中 24 头猪只体长、肩宽、臀宽的平均绝对误差均小于 3 cm,平均绝对百分比误差分别维持在 4%、6%和 7%以内.数据处理速度提升为 19.3 帧/s.[结论]本研究提出的基于关键点检测和多目标跟踪的猪只体尺估计方法为猪场生产场景提供了一个轻量化、易部署的自动体尺测量解决方案.
Abstract
[Objective]To reduce the manual workload of measuring pig body measurements in pig farms and improve measurement accuracy and efficiency.[Method]An automatic pig body measurement estimation method based on keypoint detection and multi-object tracking was proposed.The method utilized Yolov8-Pose model to identify keypoints and bounding boxes of individual pigs.ByteTrack algorithm was employed for real-time tracking of the pig herd.Regions of interest were introduced to mitigate image distortion and improve recognition speed.Additionally,a posture and anomaly detection filter algorithm was designed to reduce errors caused by motion blur,posture abnormality and other factors.[Result]The mean absolute errors of the body length,shoulder width,and hip width of 24 pigs in five pigpens were less than 3 cm,the mean absolute percentage errors were maintained below 4%,6%and 7%respectively.The data processing speed reached 19.3 frames/s.[Conclusion]The proposed method for pig body measurement estimation based on keypoint detection and multi-object tracking provides a lightweight and easily deployable solution for automatic body measurement in pig farming scenarios.
关键词
Yolov8-Pose/感兴趣区域/体尺估计/关键点/目标检测框/猪Key words
Yolov8-Pose/Region of interest/Body measurement estimation/Keypoint/Target detection box/Pig引用本文复制引用
基金项目
重庆市技术创新与应用发展专项重点项目(cstc2021jscxdxwtBX0008)
国家生猪技术创新中心先导科技项目(NCTIP-XD/B10)
出版年
2024
国家科技期刊平台
NSTL
万方数据