Chinese Scientists Develop Lightweight Model to Recognize Beef Cattle Behavior in Grassland Pastures

Beijing — Chinese scientists have successfully developed a lightweight model capable of recognizing beef cattle behavior in grassland pastures, which is expected to improve the efficiency of cattle management and feeding.

The model, named MASM-YOLO, was proposed by researchers at the Agricultural Information Institute of the Chinese Academy of Agricultural Sciences. The related research has been published in the journal Computers and Electronics in Agriculture.

Accurate and rapid identification of typical cattle behaviors is essential for disease diagnosis, estrus monitoring, calving prediction, and health assessment.

MASM-YOLO enables multi-behavior recognition under complex conditions and is suitable for real-time execution on mobile robots.

The model integrates technologies such as the Multi-Scale Focus and Extraction Network and the Adaptive Decomposition and Alignment Head to address challenges like lighting variations, motion blur, and occlusions within cattle groups.

MASM-YOLO can rapidly identify six typical behaviors of beef cattle, including feeding, resting, locomotion, and licking, striking an optimal balance between recognition accuracy and computational efficiency.

The model is expected to provide critical technical support for the full-scale development of grazing robots.