40 / 2025-03-31 18:00:19

A health Evaluation Method for Laying Hens Based on Multi-objective Behavioral Tracking

Behavioral Detection；Multi-objective Tracking；Health of Laying Hens

3. 智慧牧场与动物福利 > （1）智慧养殖的利惠：提高动物福利与健康、环境、经济效益与劳动力

With the rapid development of smart agriculture, higher requirements have been put forward for individual health monitoring and productivity assessment during large-scale egg farming. Traditional unhealthy laying hens identification relies on manual observation and RFID tracking, which has the limitations of low efficiency, high stress risk and inability to quantify behavioral characteristics. Aiming at the above problems, this paper proposes a multi-objective behavioral tracking-based health assessment model for laying hens. Firstly, an improved YOLOv11n behavioral recognition model is constructed, the SPPF module is reconfigured in the backbone layer to compress the computational redundancy, and the CBAM attention mechanism is embedded in the detection head to strengthen the capturing of key behavioral features, such as feeding/drinking. On the self-constructed laying hen behavioral dataset containing 4800 annotated data, the model improves the recognition accuracy of feeding (98.3%) and drinking (97.8%) by 3.2-3.7 percentage points compared with that of YOLOv11n. Secondly, it improves Bytetrack multi-target tracking algorithm, and enhances the MOTA in the occlusion scenario to 91.5% and the HOTA indicator to 80.0% through reconfiguring the Kalman Filtered state vectors. The HOTA index reaches 80.2%; finally, we construct a health evaluation model for laying hens based on behavioral data analysis, selecting the key time period of 3 minutes after feeding, fusing 12-dimensional behavioral features, and training the logistic regression classifier through feature standardization processing and LBFGS optimizer, which achieves an accuracy rate of 96.8% on the test set of 347 laying hens, which is significantly higher than the models of SVM (85.3%) and LightGBM ( 91.2%) and other models significantly. This method can provide reliable decision support for accurate culling of unhealthy chickens in farms, which improves the accuracy and efficiency of egg farming management.