The Body Condition Score (BCS) is a key indicator of dairy cattle’s health, production efficiency, and environmental impact. Manual BCS assessment is subjective and time-consuming, limiting its scalability in precision agriculture. This study utilizes computer vision to automatically assess cattle body condition by analyzing tail features, categorizing BCS into five levels (3.25, 3.50, 3.75, 4.0, 4.25). SE attention improves feature selection by adjusting channel importance, while spatial attention enhances spatial information processing by focusing on key image regions. EfficientNet-B0, enhanced by SE and spatial attention mechanisms, improves feature extraction and localization. To facilitate edge device deployment, model distillation reduces the size from 23.8 MB to 8.7 MB, improving inference speed and storage efficiency. After distillation, the model achieved 91.10% accuracy, 91.14% precision, 91.10% recall, and 91.10% F1 score. The accuracy increased to 97.57% for ±0.25 BCS error and 99.72% for ±0.5 error. This model saves space and meets real-time monitoring requirements, making it suitable for edge devices with limited resources. This research provides an efficient, scalable method for automated livestock health monitoring, supporting intelligent animal husbandry development.

Accurate monitoring of cattle behavioral time budgets is crucial for early disease detection and welfare assessment. Changes in durations of standing, lying, and eating are known to be early indicators of health issues such as lameness and metabolic disorders. To enable low-cost, non-invasive, and real-time monitoring, this study proposes a lightweight cattle behavior recognition method based on an improved YOLO11n architecture. The model enhances multi-scale feature integration through a generalized efficient layer aggregation network (GELAN), improves feature extraction via a multidimensional collaborative attention (MCA) mechanism, and achieves efficient cross-scale fusion using a bidirectional feature pyramid network (BiFPN). Depthwise separable convolution (DWConv) is incorporated to reduce computational load. Experimental results demonstrate high recognition accuracy, with mAP@0.5 values of 91.2%, 91.0%, and 93.9% for standing, lying, and eating, respectively. The model was subsequently compressed using a Layer-adaptive Magnitude-based Pruning (LAMP) algorithm, resulting in a final model of only 1.06 × 10⁶ parameters, a computational cost of 6.3 GFLOPS, and a weight size of 2.4 MB, while retaining 90.7% mAP@0.5. This highly efficient system is suitable for deployment on resource-constrained edge devices, providing a practical tool for continuous cattle monitoring. It offers a viable pathway for farmers to adopt precision livestock farming practices, facilitating early health intervention and promoting animal welfare.