An IoT Edge-Intelligent Estrus Detection Method for Dairy Cows Using an Improved YOLOv11n Model

Accurate identification of the estrus period in dairy cows is crucial for ensuring stable milk production and promoting intelligent livestock management. Traditional approaches relying on manual observation or single-sensor detection often suffer from low efficiency and poor real-time performance. To address these limitations, this paper proposes an IoT edge-intelligent estrus detection method for dairy cows based on an improved YOLOv11n model, which can be integrated into front-end cameras to enable real-time, contactless monitoring of estrus behavior. The proposed model recognizes the characteristic “mounting behavior” exhibited during estrus and incorporates three structural enhancements into the YOLOv11n framework: the SE-LDown downsampling module to enhance small-object detection, the SimSPPF spatial pyramid pooling module to accelerate inference, and the TransHead detection head employing an attention mechanism to improve recognition accuracy under complex conditions. Experimental results demonstrate that the improved model achieves a detection accuracy of 97.7% and an mAP50 of 95.1%, significantly reducing false and missed detections while maintaining high inference speed. Moreover, the model shows strong potential for deployment on edge devices within IoT systems, providing high practicality and scalability for intelligent dairy farming applications.