Multi-Class Brain Tumor MRI Images for Object Detection Training

A repository of MRI images organized for brain tumor object detection, containing four primary classes: glioma tumor, meningioma tumor, pituitary tumor, and no tumor. The dataset includes separate train, validation, and test subsets with images and labels, and provides pre-trained model weights for multiple detection frameworks. It was authored by Md. Sadman Haque and last updated in May 2026.

Use Cases

• Training object detection models for brain tumor localization based on the provided annotated MRI images.
• Benchmarking performance of different detection architectures like YOLO or DETR based on the included pre-trained weights.
• Evaluating model generalization on a multi-class tumor classification task using the structured train/valid/test splits.
• Reproducing research in medical AI due to the organized repository structure supporting efficient training and evaluation.

Strengths

• Includes 1.5 GB of MRI image data organized for object detection.
• Provides pre-trained model weights for multiple frameworks including Detectron2, DETR, and several YOLO variants.
• Repository is structured with clear train/validation/test splits and configuration files to support reproducibility.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• The no_tumor category contains only images without annotation labels, which may limit its use for some detection tasks.

Provenance

Source

figshare

Freshness

Last updated 2026-05-31 08:33:11; freshness should be verified.