Comparative Analysis Of YOLO And Transformer Based Architectures For Animal X-Ray Anatomy Recognition
DOI:
https://doi.org/10.29284/a517e457Keywords:
.Abstract
Accurate interpretation of veterinary radiographs is essential for diagnosis, treatment planning, and surgical decision-making in animal healthcare. However, manual analysis of X-ray images is time-consuming and re- quires specialized expertise. Recent advances in deep learning have demonstrated significant potential in automating medical image analysis; however, research in veterinary imaging—especially in the classification of anatomical body parts in animal radiographs—remains limited. This paper presents comparison of popular deep learning models, namely , YOLOv8, YOLOv11, Vision Transformer (ViT), and ResNet50, for automatically identifying body parts in dog X-rays. The models were trained and tested on a carefully prepared dataset of 453 dog X-ray images, covering five body regions: skull, chest, abdomen, front leg, and back leg. Models were assessed on accuracy, precision, recall, F1-score, and inference speed. Experimental results demonstrate that YOLO-based models achieve competitive accuracy while offering significantly faster inference compared to ViT and ResNet50, making them suitable for real-time veterinary applications. The Vision Trans- former achieved strong classification performance on limited data due to its attention-based feature learning, while ResNet50 served as a reliable baseline. The findings presented in this paper can support veterinary diagnostic systems and assist practitioners in clinical workflows.
References
[1] T. Banzato et al., “Automatic classification of canine thoracic radiographs using deep learning,” Scientific Reports, vol. 11, Art. no. 3964, 2021.
[2] T. Banzato et al., “AI-based algorithm for the automatic evaluation of image quality in canine thoracic radiographs,” Scientific Reports, 2023.
[3] J. Dumortier et al., “Deep learning in feline thoracic radiographs: A CNN approach for pulmonary abnormality detection,” Scientific Reports, 2022.
[4] T. Banzato et al., “Interspecies and self-supervised transfer learning for veterinary thoracic radiographs,” Scientific Reports, 2023.
[5] J. Park et al., “Development of a deep learning model for automatic detection of narrowed intervertebral disc space sites in caudal thoracic and lumbar X-ray images of dogs,” Frontiers in Veterinary Science, vol. 11, Art. no. 1453765, 2024.
[6] J. Park et al., “Detection of spondylosis deformans in thoracolumbar and lumbar lateral X-ray images of dogs using a deep learning network,” Frontiers in Veterinary Science, 2024.
[7] C. A. Duenas et al., “Radiographic dataset for VHS determination learning process,” Mendeley Data, 2020.
[8] J. Redmon et al., “You only look once: Unified, real-time object detection,” arXiv preprint arXiv:1506.02640, 2016.
[9] N. H. Harun and N. Fared, “An experiment on lung disease classification using YOLOv8,” Applied Mathematics and Computational Intelligence, vol. 13, no. 3, pp. 176–185, 2024.
[10] A. Sharma et al., “Transfer learning for wildlife classification: Evaluating YOLOv8 against DenseNet, ResNet, and VGGNet,” Artificial Intelligence and Cloud Networking Journal, vol. 6, no. 4, 2024.
[11] Y. Li et al., “Deep learning-based detection of veterinary disease using YOLOv8,” EAI Endorsed Transactions on Pervasive Health and Technology, vol. 9, no. 36, 2024. [Online]. Available: https://publications.eai.eu/index.php/phat/article/view/5577
[12] A. Dosovitskiy et al., “An image is worth 16×16 words: Transformers for image recognition at scale,” arXiv preprint arXiv:2010.11929, 2020.
[13] A. Zimbalist et al., “Detecting bone lesions in X-ray under diverse acquisition conditions,” Scientific Reports, Nature Publishing Group, 2023. [Online]. Available: https://www.nature.com/articles/s41598-023-46345-z
[14] H. Touvron et al., “Training data-efficient image
transformers and distillation through attention,” arXiv preprint arXiv:2212.07792, 2022.
[15] K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” arXiv preprint arXiv:1512.03385, 2015.
[16] N. Deekonda, “Assessing cardiomegaly in dogs using a simple CNN model,” arXiv preprint, 2024.
[17] A. Dourson et al., “Vet-DINO: Learning anatomical understanding through multi-view distillation in veterinary imaging,” arXiv preprint arXiv:2505.15248, 2025.
[18] G. Jocher, A. Chaurasia, and J. Qiu, "YOLOv8: Ultralytics Next-Generation Object Detection Model," Ultralytics, 2023. [Online]. Available: https://github.com/ultralytics/ultralytics
Downloads
Published
Issue
Section
License
This work is licensed under a Creative Commons Attribution 4.0 International License.
This work is licensed under a Creative Commons Attribution 4.0 International License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
