Implementation of the YOLOv8n Model for Automatic Owl Detection in Swiftlet Farming Buildings

Putra, Iqbal Kurniawan Asmar
Apriska Prameswari
Fikri, Muhammad Ainul
Suhari, Ahmad Riznandi

Journal of Advanced in Information and Industrial Technology (JAIIT)
Vol. 7 No. 2 (2025): Nov

Object detection based on digital images is a rapidly developing field in the application of intelligent systems. This study aims to create an automatic owl detection system utilizing the YOLOv8 deep learning model as a pest mitigation measure in the swiftlet farming industry. Owls are known to enter swiftlet houses at night and prey on the birds, causing economic losses. Owl image datasets were obtained from the Roboflow platform and annotated in YOLO format. The model was trained using the YOLOv8-nano architecture with a 640×640 pixel input resolution. The evaluation results showed that the model achieved a mAP@0.5 of 96.82% and mAP@0.5:0.95 of 70.5%, with a precision of 97.2% and a recall of 93.38%. These results indicate that the YOLOv8 model performs well and has the potential to be implemented as an automatic monitoring system in swiftlet farming environments.