一种基于像素级活动量的生猪健康检测方法及系统

By using a pixel-level activity-based method for pig health detection, a dense optical flow field is generated using the RAFT deep learning optical flow model. The activity intensity of each pixel is calculated and a heat map is generated. This solves the problems of individual tracking failure and insufficient detection of subtle activity changes in existing technologies in intensive farming scenarios, and achieves high-sensitivity detection and accurate localization of early health abnormalities.

CN122091213BActive Publication Date: 2026-07-17SOUTH CHINA AGRICULTURAL UNIVERSITY
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
SOUTH CHINA AGRICULTURAL UNIVERSITY
Filing Date
2026-04-21
Publication Date
2026-07-17

AI Technical Summary

Technical Problem

Existing computer vision-based methods for detecting pig health cannot work stably in intensive farming scenarios. They cannot distinguish the source of movement, locate individuals with low activity levels, and are not sensitive to subtle health abnormalities. They also have a heavy computational burden and are difficult to achieve early warning.

Method used

A pig health detection method based on pixel-level activity was adopted. A dense optical flow field was generated by the RAFT deep learning optical flow model, the instantaneous activity intensity of each pixel was calculated, and the ROI mask was used for correction to generate an activity cumulative heat map for herd health assessment and individual anomaly localization.

Benefits of technology

It achieves stable individual localization in densely occluded environments, has extremely high sensitivity for early health anomaly detection, strong environmental adaptability and computational efficiency, meets the needs of large-scale real-time processing, and provides integrated panoramic decision support for groups and individuals.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122091213B_ABST
    Figure CN122091213B_ABST
Patent Text Reader

Abstract

本发明属于生猪健康检测技术领域，公开了一种基于像素级活动量的生猪健康检测方法及系统，方法包括：对目标猪栏的原始视频流进行数据标准化及背景建模处理，生成视频帧序列及其对应的ROI掩码；通过RAFT深度学习光流模型对视频帧序列进行高质量运动感知，生成稠密光流场；基于稠密光流场，计算每个像素点的瞬时活动强度，并结合ROI掩码对其修正，对每个像素点修正后的瞬时活动强度进行非线性时序累加，生成活动量累积热力图；基于活动量累积热力图，分别进行群体健康评估及个体异常定位，并将群体及个体异常预警可视化输出，实现生猪健康检测。本发明提供了一种计算高效、对遮挡鲁棒、能同时服务于群体异常预警和个体精确定位的生猪健康检测新范式。
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Monitoring video enhancement method for farm

    CN120634901A

  • Pig behavior-based pig health condition analysis method and system

    CN120975553A