Cattle monitoring can rely mainly on manual inspections in the fence farming. Once the cattle escaped near the edge of the fence, the conventional regulatory system could not fully detect and intervene in time, resulting in the loss of cattle. In this study, a cattle area monitoring and boundary-crossing early warning system was proposed for the cattle using fence-type breeding. Instance segmentation and multi-object tracking, named STWBD (Segment Tracking with Boundary Detection), were also utilized to monitor the cattle activities in real time within the fenced area. Intelligent early warnings of the potential boundary-crossing behaviors were then provided to enhance the safety level of the fenced cattle farming. Firstly, an improved RDL-YOLO11n-seg instance segmentation was constructed into the framework. The C3K2-RetBlock module was also introduced to the original YOLO11n-seg, in order to enhance the response capability to the spatial displacement of the cattle. The C2DA structure was combined to improve the expression of the multi-scale features. The Segment-LSCD module was employed to optimize the segmentation head. The object detection and instance segmentation were facilitated to accurately capture the cattle contours and postures. As a result, the spatial perception and real-time monitoring of the cattle were enhanced within the fenced area. Secondly, the DeepOCSORT algorithm was assigned an identity ID to each cattle, particularly for the continuous tracking in different scenarios. Some challenges were effectively handled, such as the small target scale, severe occlusion, low lighting, and cattle close to the boundary. The continuity and stability of the multi-object tracking were provided for the reliable trajectory data to detect the later boundary-crossing behavior. In terms of the fence boundary design, the actual structure of the cattle shed fences was combined to construct a polygonal electronic fence. The safe activity area of the cattle was defined for the continuous analysis of the movement trajectories and position distribution of the cattle. The system was integrated with some functions, such as the trajectory visualization, real-time speed tracking, and statistics of the cattle population. The STWBD was used to monitor the boundary-crossing behavior of the cattle in real time and promptly trigger the intelligent early warnings. Experimental results showed that the precision, recall, and segmentation accuracy of the RDL-YOLO11n-seg model reached 94.9%, 87.4%, and 92.5%, respectively, which were 2.0, 3.1, and 3.1 percentage points higher than the original, and the parameters were reduced by 17.8%, with a floating-point operation count of 8.6 G. The high-order tracking accuracy (HOTA), multiple object tracking accuracy (MOTA), and identity F1-Score (IDFI) of DeepOCSORT were 81.4%, 91.6%, and 95.8%, respectively, while the precision, recall, and F1 score of the escape event detection were 96.2 %, 100.0 %, and 98.1 %, respectively. The movement trajectories of the cattle were efficiently monitored in the small-scale fenced farming environments. The boundary-crossing escape events were effectively identified to provide the reliable technical support for the intelligent early warning and supervision of the cattle. This finding can provide real-time access to the cattle population, distribution, and movement patterns. While the risk was predicted after trajectory and speed analysis. The finding can offer a scientific basis for the secure and efficient management of the confined cattle farming.

Behavior recognition of beef cattle has been one of the most essential means in the intelligent breeding and health monitoring of cattle sheds. Its recognition often depended on the breeding environment, including the lighting conditions, stocking density, and ground conditions. However, the behavioral identification can be challenging due to the blurred boundary of individual cattle. The beef cattle, as a group animal, can usually tend to rest in groups of 3 to 5. The behavioral patterns and movement of the cattle can interfere with the accuracy of behavior recognition. Especially, the image visibility is significantly reduced during monitoring at night. In addition, the noise interference can further reduce the performance of the traditional behavior recognition under night breeding environments. Conventional recognition of beef cattle behavior can also be limited to the low image visibility, fuzzy and variable behavior, as well as the noise interference. In this study, a framework of recognition was proposed for the beef cattle at night using low-light image enhancement. Firstly, the Lighten Diffusion model was used to enhance the behavior image of beef cattle at night. The clarity and visibility of the image enhanced the target detection. Secondly, the YOLOv11n model was utilized to construct the C3k2-CAFormerCGLU module. Local important features of beef cattle were obtained using the gating mechanism. The noise and irrelevant information were suppressed to avoid misjudgment on the individual boundary of beef cattle. The beef cattle behaviors were effectively distinguished against the background noise or cattle overlapping area. The spatial-to-depth conversion convolution (SPDConv) was used to improve the original subsampling of the model. The spatial and depth information flow was optimized in the convolution operation. There was a more accurate feature expression of the beef cattle at a long distance. The recognition accuracy of the beef cattle behavior was improved from a far perspective. Finally, a Feature Focusing module was introduced into the neck Network. The feature focusing pyramid network (FFPN) was constructed to enhance the feature expression of the target region. The fine-grained feature was extracted under different receptive fields. At the same time, a cross-scale feature fusion was adopted to make the features with rich context information, in order to propagate effectively between different detection scales. Furthermore, the multi-scale target recognition was improved in complex environments. The experimental results showed that the image enhancement was achieved in the four unsupervised low-light enhancement algorithms, including Zero-DCE, Zero-DCE++, Enlighten GAN, and Lighten Diffusion, compared with the low-light image enhancement dataset. The Lighten Diffusion algorithm performed best in the low-light image enhancement tasks, with the highest peak signal to noise ratio (19.79), the highest structural similarity index measure (91.7%), and the lowest mean squared error (691.46). The dataset of beef cattle behavior before and after the enhancement of the Lighten Diffusion was sent into the YOLOv11n target detection model for training. The average recognition accuracy of beef cattle behavior increased by 2.5 percentage points after the enhancement. The precision, recall, and mAP@0.5 of the improved CSF-YOLOv11n model reached 93.9, 87.3, and 94.3, respectively, in terms of the target detection. Compared with the Faster-RCNN, RT-DETR, YOLOv5n, YOLOv7, YOLOv8n, YOLOv9t, YOLOv10n, and YOLOv11n, the mAP@0.5 of the model increased by 8.4, 7.4, 5.9, 6.4, 5.6, 6.4, 5.7, and 4.9 percentage points, respectively. The finding can also provide a strong reference to realize the healthy breeding of beef cattle under all weather conditions.