BrainVision - Test Data

This directory contains a small balanced test subset used for evaluating the BrainVision ONNX model with real MRI images.

The images are sourced from the following public dataset:

Kaggle Dataset:

• https://www.kaggle.com/datasets/sartajbhuvaji/brain-tumor-classification-mri

Folder Structure

Each folder contains 20 randomly selected MRI images from its category:

test-data/
├── Glioma/
├── Meningioma/
├── No Tumor/
└── Pituitary/

This subset is designed to provide a fast and reliable way to:

• Test the browser-based classifier
• Validate ONNX inference
• Debug preprocessing steps
• Compare predictions with original labels
• Demonstrate model performance in documentation or demos

---

Purpose

This test set is only for evaluation, not training. It helps ensure model and UI behavior is correct in real-world cases.

---

Licensing

All images originate from the Kaggle dataset listed above. Redistribution should follow the dataset's licensing terms.

---

Notes

• Images are unmodified except for random selection.
• No resizing, filtering, or preprocessing was applied.
• Ideal for manual and automated testing in both development and documentation.
• The model may make incorrect predictions --- this is expected in any real-world machine-learning system.
• Correctness of the image labels belongs to the dataset creators, not to this project.