README.md

---
title: Tuberculosis Detection ViT
emoji: 🩺
colorFrom: blue
colorTo: green
sdk: gradio
sdk_version: 4.36.1
app_file: app.py
pinned: false
short_description: Classify chest X-ray images as Normal or Tuberculosis
tags:
  - medical-imaging
  - tuberculosis-detection
  - vision-transformer
  - pytorch
  - gradio
---

# 🩺 Tuberculosis Detection with Vision Transformer

Classify chest X-ray images as **Normal** or **Tuberculosis** using a Vision Transformer (ViT) model.

## How to Use (Web Interface)
- Upload a chest X-ray image (JPEG/PNG).
- Click "Predict".
- View the prediction, confidence score, and probabilities.

## How to Use (API)

### Standard Endpoint
**URL**: `https://sukhmani1303-tuberculosis-vit-model.hf.space/api/predict/`
**Method**: POST
**Content-Type**: multipart/form-data
**Input**: Image file (JPEG/PNG)
**Output**: JSON response

```python
import requests

url = "https://sukhmani1303-tuberculosis-vit-model.hf.space/api/predict/"
files = {"file": open("chest_xray.jpg", "rb")}
response = requests.post(url, files=files)
print(response.json())
```

### Raw Debug Endpoint
**URL**: `https://sukhmani1303-tuberculosis-vit-model.hf.space/api/predict_raw/`
**Method**: POST
**Content-Type**: multipart/form-data
**Input**: Image file (JPEG/PNG)
**Output**: JSON response with raw debug information

```python
import requests

url = "https://sukhmani1303-tuberculosis-vit-model.hf.space/api/predict_raw/"
files = {"file": open("chest_xray.jpg", "rb")}
response = requests.post(url, files=files)
print(response.json())
```

## Expected Response Format

```json
{
  "status": "success",
  "prediction": "Normal",
  "confidence": 0.8542,
  "probabilities": {
    "Normal": 0.8542,
    "Tuberculosis": 0.1458
  }
}
```

## Medical Disclaimer
This tool is for **educational and research purposes only**. It is not intended for medical diagnosis. Always consult qualified healthcare professionals for medical advice and diagnosis.

## Model Information
- **Architecture**: Vision Transformer (ViT)
- **Task**: Binary classification (Normal vs Tuberculosis)
- **Input**: Chest X-ray images
- **Image Size**: 224x224 pixels