Skin Burn Image Classification Labels by Burn Degree

358 retrospective images from Karadeniz Technical University Farabi Hospital were combined with public datasets from Roboflow Universe and Kaggle. All images were labeled into four burn degrees by expert general surgeons. The dataset was created by Uğur Şevik and last updated in March 2026.

Use Cases

• Training object detection models for burn classification based on labeled burn degrees.
• Benchmarking YOLO-based segmentation architectures for medical image analysis.
• Developing decision support tools for clinical burn assessment based on expert-labeled images.

Strengths

• Dataset combines 358 proprietary images with large public datasets, likely increasing diversity.
• Labels were applied by expert general surgeons, suggesting clinical relevance.
• Data augmentation and preprocessing techniques were applied, which may enhance model generalizability.

Limitations

• The dataset is very small at 5.5 KB, indicating it likely contains only summary statistics or labels, not the images themselves.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

Combination of a proprietary collection from Karadeniz Technical University Farabi Hospital and public datasets from Roboflow Universe and Kaggle.

Collection Method

Retrospective image collection and expert labeling.

Freshness

Last updated 2026-03-18 17:28:38; freshness should be verified.