Identification-Based Cattle Tracking Using Feature Matching Models

In this study, a method is proposed for the identification and tracking of cattle in a barn environment. Traditional tracking algorithms fail when cattle leave the camera frame. Therefore, a feature matching-based approach was developed to address this problem. First, YOLO11x model were used to detect the cattle, which were then cropped and saved. Next, keypoints for each cow were extracted using the SuperPoint model. Finally, the matching of cattle appearing in different frames and time periods was performed using the LightGlue model. The results of this study demonstrate that the proposed method enables cattle to be recognized and matched across different angles and time frames.