Surgical Instrument Images for Multiview Object Detection

The 'Surgical Tool Recognition Full Multiview' dataset contains 2,520 images of individual surgical instruments for object detection tasks. It includes 2,016 training, 252 validation, and 252 test images across five instrument classes. Author joonhaim exported the dataset to Hugging Face in April 2026.

Use Cases

• Train an object detector like YOLO to classify surgical instruments (clamp, needle_holder, scalpel, shear, tweezer) from multiview images.
• Benchmark model performance on the defined validation and test splits of 252 images each.
• Develop a tool presence detection system for surgical workflow analysis using the labeled image data.
• Fine-tune a pre-trained vision model on the 2,016 training images for domain-specific surgical tool recognition.

Strengths

• 2,520 total images provide a foundation for model training.
• Five distinct instrument classes (clamp, needle_holder, scalpel, shear, tweezer) cover common tools.
• Pre-defined train/validation/test splits (2016/252/252) facilitate standardized evaluation.

Limitations

• The dataset size of 2,520 images is modest for training deep learning models from scratch.
• Limited to five tool classes, which may not represent the full diversity of surgical instruments.
• No information on image resolution, lighting conditions, or potential class imbalance.

Provenance

Source

Author joonhaim on Hugging Face.

Collection Method

Originally created in YOLO format and exported to a Hugging Face-friendly structure.

Freshness

Last updated on the platform in April 2026.