4,083 musculoskeletal X-ray images annotated for fracture classification, localization, and segmentation. The collection includes multi-format annotations such as bounding boxes and segmentation masks for bone fracture analysis.
Use Cases
- Train a fracture localization model using the provided bbox coordinates
- Develop a pixel-level fracture segmentation tool using the segmentation mask data
- Implement a classification system using the fracture classification labels to detect the presence of fractures in radiographs
Strengths
- 4,083 X-ray images specifically targeting bone fractures
- Includes both bounding box (bbox) and segmentation mask annotations
- Distributed under the CC-BY 4.0 license for open access
- Supports multi-task learning across classification, localization, and segmentation