Controlled comparative study of YOLOv8-Pose, YOLOv11-Pose, and Detectron2 for vertebrae detection and keypoint estimation
PLOS One · 21.04.2026 · 21 (4) · e0347290
DOI: 10.1371/journal.pone.0347290 ↗Kurz gefasst
Vergleichsstudie in der Zeitschrift PLOS One, online erschienen am 21.04.2026. Verfasst von Avinash Bansal und 2 weiteren Autor:innen. Beteiligte Einrichtung: Department of Cybernetics and Biomedical Engineering, VSB-Technical University of Ostrava, FEECS, Ostrava, Poruba, Czech Republic. Hauptthemen laut MEDLINE: Spine, Image Processing, Computer-Assisted. Die Arbeit nennt 43 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: European Union under the LERCO Project, Biomedical Engineering Systems XXI Project.
Abstract (Originalsprache)
Accurate vertebrae detection with precise keypoint localization is essential for medical image analysis and clinical applications, such as spinal alignment assessment. This study presents a controlled, task-driven comparison of four pose-based deep learning models for vertebrae detection and keypoint estimation: YOLOv8n-Pose, YOLOv11n-Pose, and Detectron2 (Keypoint R-CNN) with ResNet-50 and ResNet-101 backbones. A single-class vertebrae dataset with bounding box annotations and four anatomically meaningful keypoints per vertebra was evaluated under consistent training settings. Performance was assessed in terms of keypoint localization accuracy, detection precision, inference speed, and vertebrae-specific prediction behavior, including detection completeness and duplicate detections, which are critical for clinical usability. Results show that YOLOv11n-Pose provides the best balance between keypoint accuracy and inference efficiency, while YOLOv8n-Pose achieves the fastest inference with competitive performance. Detectron2-based models exhibit lower keypoint accuracy, slower inference, and frequent duplicate predictions that obscure correct anatomical landmarks. Additional experiments indicate that larger YOLO variants (particularly YOLOv8l-Pose) improve accuracy, albeit at increased computational cost. Overall, the results demonstrate that model capacity or architectural recency alone does not guarantee superior performance in vertebrae keypoint detection, highlighting the importance of anatomy-aware model selection for spine imaging applications.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Autor:innen und Institutionen
- Department of Cybernetics and Biomedical Engineering, VSB-Technical University of Ostrava, FEECS, Ostrava, Poruba, Czech Republic.
ORCID-Profile: Avinash Bansal
Förderung
- European Union under the LERCO Project
- Biomedical Engineering Systems XXI Project
Zitieren
APA
Bansal, A., Kubíček, J., & Garg, S. (2026). Controlled comparative study of YOLOv8-Pose, YOLOv11-Pose, and Detectron2 for vertebrae detection and keypoint estimation. PLOS One, 21(4), e0347290. https://doi.org/10.1371/journal.pone.0347290
Vancouver
Bansal A, Kubíček J, Garg S. Controlled comparative study of YOLOv8-Pose, YOLOv11-Pose, and Detectron2 for vertebrae detection and keypoint estimation. PLOS One. 2026;21(4):e0347290. doi:10.1371/journal.pone.0347290
BibTeX
@article{Bansal2026,
author = {Bansal, Avinash and Kubíček, Jan and Garg, Shalini},
title = {Controlled comparative study of YOLOv8-Pose, YOLOv11-Pose, and Detectron2 for vertebrae detection and keypoint estimation},
journal = {PLOS One},
year = {2026},
volume = {21},
number = {4},
pages = {e0347290},
doi = {10.1371/journal.pone.0347290}
}
EUVIMED