A method for identifying abnormal behavior of forest musk deer
By using an improved YOLOv8-Pose model and pressure distribution sensor array, the problems of background interference and limb joint localization in the detection of abnormal behavior of musk deer were solved, realizing accurate identification and non-invasive monitoring of abnormal behavior of musk deer, and improving detection accuracy and ease of operation.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CHENGDU UNIV OF INFORMATION TECH
- Filing Date
- 2026-05-13
- Publication Date
- 2026-07-24
AI Technical Summary
Traditional models for detecting abnormal behavior in musk deer are susceptible to background interference, resulting in large detection biases and poor accuracy in limb and joint positioning, making it difficult to support refined behavioral analysis. Furthermore, traditional bioelectric monitoring methods cause stress to musk deer and are complex to operate.
An improved YOLOv8-Pose model was adopted, embedding a lightweight CBAM attention module and a CoordConv layer, combined with a LiteACmix module and a FocalLoss loss function. The musk deer's scent expulsion behavior was monitored through a pressure distribution sensor array. The sensor was discarded, and the behavior was distinguished by the pressure center of gravity and the pressure fluctuation characteristics of the anal region.
It enables accurate identification and non-invasive monitoring of abnormal behaviors in musk deer, improves the accuracy of limb joint positioning, reduces stress on musk deer, lowers operational complexity, and meets the needs of real-time monitoring.
Smart Images

Figure CN122454640A_ABST