一种基于双流协同机制的奶牛跛行识别方法
The dairy cow lameness identification method based on a dual-stream collaborative mechanism, combined with lightweight target detection and kinematic constraints, solves the problems of robustness and computational energy consumption in complex environments in existing dairy cow lameness identification technologies, and achieves efficient and accurate dairy cow lameness monitoring.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- ZHENGZHOU UNIVERSITY OF LIGHT INDUSTRY
- Filing Date
- 2026-06-02
- Publication Date
- 2026-07-17
AI Technical Summary
Existing methods for identifying lameness in dairy cows have low robustness in complex pasture environments, high false alarm rates for local high-frequency movements, and high energy consumption for edge computing, making it difficult to meet the needs of all-weather real-time monitoring.
A method for recognizing lameness in dairy cows based on a dual-stream collaborative mechanism is adopted. This method combines lightweight target detection and multi-target tracking algorithms. The gait features of dairy cows are extracted by a lightweight 2D convolutional ResNet deep flow network with embedded TSM modules. The physical kinematic constraints of absolute trajectory displacement and trajectory expansion coefficient are combined with a conditional gating mechanism for verification, thereby achieving efficient behavior recognition.
It significantly improves the accuracy of lameness detection in dairy cows and the robustness of the system. It enables real-time and efficient processing of multiple monitoring videos on edge computing devices, reducing computational overhead and false alarm rate.
Smart Images
Figure CN122397643A_ABST