Reproduce “Evaluating Stenosis Detection with Grounding DINO, YOLO, and DINO-DETR” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Muhammad Musab Ansari
This paper evaluates three object detection models—YOLO, DINO-DETR, and Grounding DINO—for automated stenosis detection in X-ray coronary angiography images using the ARCADE dataset. The study found transformer-based models generally achieved higher mean Average Precision for this task, while YOLO offered computational efficiency and competitive performance on smaller stenosis regions.
Key insights
- Transformer-based models (DINO-DETR and Grounding DINO) generally achieved higher mean Average Precision compared to YOLO for automated stenosis detection.
- DINO-DETR prioritized precision, producing fewer but more precise detections, while Grounding DINO offered a strong balance between precision and recall, identifying more regions.
- YOLO provided a reasonable trade-off, effectively capturing anatomical structures and performing better on smaller stenosis regions while offering computational efficiency.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.