README.md

---

title: MRI Orientation
emoji: 🐠
colorFrom: yellow
colorTo: gray
sdk: gradio
sdk_version: 6.2.0
app_file: app.py
pinned: false
---

# MRI Orientation Predictor

This is a simple web application for predicting the orientation of MRI images using a YOLOv11 classification model that is already pretrained on several MRI datasets.
It classifies images into **axial**, **coaxial**, or **sagittal** orientations and provides a confidence score.

---

## Features

* Predict MRI orientation for uploaded images.
* Display confidence score for the prediction.
* Easy-to-use web interface powered by **Gradio**.
* Supports `.jpg`, `.jpeg`, and `.png` images.

---

## Model

The model is a **YOLOv11 classification model** saved as `best.pt`.
The model outputs short labels:

* `ax` → **axial**
* `co` → **coaxial**
* `sa` → **sagittal**

The application maps these short labels to full orientation names for easier understanding.

---

## Installation

Install required packages:

```bash
pip install -r requirements.txt
```

### Requirements.txt

The `requirements.txt` should include at least:

```
ultralytics
torch
torchvision
gradio
Pillow
```

---

## Usage

1. Place your YOLOv11 model file `best.pt` in the project directory.
2. Run the Gradio app:

```bash
python app.py
```

3. Upload an MRI image in the interface.
4. The predicted orientation and confidence will be displayed.

---

## Example Output

```
Orientation: axial | Confidence: 0.95
Orientation: coaxial | Confidence: 0.87
Orientation: sagittal | Confidence: 0.92
```

---

## Notes

* Make sure your model is a YOLOv11 **classification model** saved as `.pt`.
* The model should return labels `"ax"`, `"co"`, `"sa"` which are mapped to full names.
* Confidence is reported as a float between 0 and 1.

---