---
title: reCAPTCHA Solver 3x3
emoji: 🤖
colorFrom: blue
colorTo: green
sdk: docker
pinned: false
license: mit
---

# 🤖 reCAPTCHA 3x3 Detection API

Solves 3x3 grid reCAPTCHA challenges using YOLO object detection.

## 🎯 Features

- ⚡ Fast inference (2-3s per request)
- 🎯 High accuracy with YOLO detection
- 🔄 Auto-lowering confidence for better detection
- 📊 Built-in metrics and monitoring
- 🌐 CORS enabled for browser extensions

## 📡 API Endpoints

### POST /predict

Predict which tiles to click in a 3x3 grid.

**Request:**
```json
{
  "image": "data:image/png;base64,iVBOR...",
  "challenge_title": "crosswalks"
}
```

**Response:**
```json
{
  "success": true,
  "tiles_to_click": [2, 5, 8],
  "num_detections": 12,
  "confidence_used": 0.20,
  "latency_s": 2.341
}
```

### GET /health

Check API health and statistics.

**Response:**
```json
{
  "status": "healthy",
  "model_loaded": true,
  "requests_total": 42,
  "requests_successful": 40,
  "avg_latency_s": 2.5
}
```

## 🚀 Usage

### From Browser Extension

```javascript
const response = await fetch('https://YOUR-SPACE.hf.space/predict', {
    method: 'POST',
    headers: {'Content-Type': 'application/json'},
    body: JSON.stringify({
        image: screenshotBase64,
        challenge_title: 'crosswalks'
    })
});

const result = await response.json();
console.log('Tiles to click:', result.tiles_to_click);
```

### From Python

```python
import requests
import base64

with open('screenshot.png', 'rb') as f:
    image_b64 = base64.b64encode(f.read()).decode()

response = requests.post('https://YOUR-SPACE.hf.space/predict', json={
    'image': f'data:image/png;base64,{image_b64}',
    'challenge_title': 'crosswalks'
})

print(response.json())
```

## 🔧 Model Details

- **Architecture:** YOLOv8 Detection
- **Input:** 640x640 RGB image
- **Output:** Bounding boxes with confidence scores
- **Classes:** Multiple object types (vehicles, crosswalks, traffic lights, etc.)

## 📊 Performance

- **Cold start:** ~15 seconds (first request)
- **Warm inference:** 2-3 seconds per request
- **Memory usage:** ~1.5GB
- **Concurrent requests:** 2-3 simultaneous

## 🎯 Tile Mapping

The API maps detected objects to a 3x3 grid:

```
[0] [1] [2]
[3] [4] [5]
[6] [7] [8]
```

Objects are mapped based on their center point within the grid.

## 🛡️ Rate Limits

- **Free tier:** 100 requests per hour
- **Timeout:** 30 seconds per request
- **Max image size:** 10MB

## 📝 License

MIT License - See LICENSE file for details

## 🔗 Related

- [4x4 Segmentation Space](https://huggingface.co/spaces/YOUR-USERNAME/recaptcha-solver-4x4)
- [Browser Extension](https://github.com/YOUR-REPO)

---

**Note:** This API is for educational and research purposes only. Use responsibly and respect website terms of service.