BIOMETRIC IDENTIFICATION OF WATER BUFFALO USING REAL-TIME FACE RECOGNITION Authors: Niyazi Hayrullah Tuvay, Orhan Ermetin, Tolga Hayit Journal: Journal of Animal and Plant Sciences (JAPS) ISSN: 1018-7081 (Print), 2309-8694 (Online) Volume: 36 Issue: 6 Year: 2026 DOI: https://doi.org/10.36899/JAPS.2026.6.0130 URL: https://doi.org/https://doi.org/10.36899/JAPS.2026.6.0130 Publisher: Pakistan Agricultural Scientists Forum Abstract:

Face recognition is increasingly used for biometric identification in both humans and animals, offering a non-invasive option for managing hard-to-handle species such as water buffalo. We present a real-time face recognition system based on YOLOv5 to accurately identify individual buffalo in a livestock setting. Our goal is to support precision livestock farming with an efficient and scalable monitoring solution. We also introduce Buffalo-22, a dataset of 4000 augmented face images from eight water buffalo collected at a breeding farm in Yozgat, Turkey. We compare our system with a traditional approach based on Local Binary Patterns in the HSV color space (LBP-HSV) to highlight the advantages of deep learning in challenging agricultural environments. The proposed YOLOv5 model achieved 99.3% average precision at IoU 0.5 (mAP@0.5), while the LBP-HSV model reached 87.8% test accuracy. These results show that convolutional neural networks can deliver accurate, real-time face recognition for water buffalo. The reported performance indicates that the proposed system is a practical tool for automatic and reliable identification on livestock farms.

Keywords: Biometric identification, face recognition, water buffalo, deep learning, YOLOv5