Cattle detection model based on improved YOLOv8

As the most dynamic and potential part of animal husbandry, cattle farming accounts for an increasing proportion of animal husbandry year by year. However, relying on the traditional artificial breeding management mode will seriously restrict the development process of large-scale cattle breeding. As big data continues to advance, the utilization of deep learning in object detection has reached a more advanced stage. Therefore, this paper improves YOLOv8 by introducing an improved lightweight Spatial Pyramid Pooling Fast Cross Stage Partial Channel module(SPPFCSPC-P) and the Feature Enhancement Module(FEM) in order to improve the model's ability to recognize and process feature maps. At the same time, the loss function of YOLOv8 is replaced by the Weighted Intersection over Union loss (WIoU) to enhance the model's performance in handling occluded targets. In this paper, the test results are verified by the Public dairy cow detection datasets, and the results show that the improved YOLOv8 model improves the Precision, Recall and Mean Average Precision (mAP) of a single cow detection task by 2.1%, 2.1% and 1.1%, respectively, compared with the baseline model.