YOLOv8-Based Object Detection for Automated Cattle Sex Identification in Tomohon City

Abstract

Accurate identification of cattle sex is essential in livestock management to support proper data recording and strategic decision-making. Conventional identification methods are often time-consuming and dependent on human expertise, which may reduce efficiency and accuracy. Therefore, this study aims to develop an automated cattle sex identification system using an object detection approach based on the YOLOv8 algorithm. The dataset consisted of primary images collected directly in Tomohon City using a smartphone camera and secondary images obtained from the Roboflow platform. All images were annotated with bounding boxes and classified into two categories: male and female cattle. The dataset was divided into training, validation, and testing sets with a ratio of 70:20:10. Model training and evaluation were conducted using the Roboflow platform, and the final model was integrated into a web-based system to enable real-time detection. The experimental results show that the proposed model achieved 97.1% precision, 92.5% recall, and 98.7% mean average precision (mAP). These findings indicate that the system performs reliably and can serve as an effective tool to support livestock data management in Tomohon City.