With the modernization and intelligence of agricultural production, the demand for animal management and monitoring is also increasing. Ostriches are an important livestock resource that requires effective monitoring and identification in breeding and management. This article proposes an improved YOLOv5s based ostrich detection model by establishing a self-made ostrich detection dataset and manually annotating it. The model introduces a lightweight network GhostNetV2, integrates position attention mechanism, uses collaborative coordinate convolution module, and replaces the loss function with MPDIoU to achieve feature extraction, reduce the number of parameters, and computational complexity Comprehensive optimization in enhancing feature fusion capability and improving the quality of target detection results. At the same time, the self-made dataset also ensures the effectiveness of the model in practical application scenarios, providing a solid theoretical foundation for subsequent ostrich breeding management, early disease warning, and agricultural intelligent development.