README.md

---
title: Dog Breed Classification API
emoji: 🐕
colorFrom: blue
colorTo: green
sdk: docker
app_port: 7860
pinned: false
---

# 🐕 Dog Breed Classification API

ConvNextV2-large-DogBreed model ile köpek ırkı tahmini yapan API.

## 🚀 Kullanım

### Endpoint

```
POST /predict_pet
```

### Request

**Content-Type:** `multipart/form-data`

**Body:**
- `image` (file): Köpek fotoğrafı (JPEG, PNG, WebP)

### Response

```json
{
  "breed": "Doberman_pinscher",
  "confidence": 0.533,
  "top_5": [
    {"breed": "Doberman_pinscher", "confidence": 0.533},
    {"breed": "Beauceron", "confidence": 0.065},
    {"breed": "German_pinscher", "confidence": 0.041},
    {"breed": "Black_and_tan_coonhound", "confidence": 0.023},
    {"breed": "Greater_swiss_mountain_dog", "confidence": 0.011}
  ],
  "model": "ConvNextV2-large-DogBreed",
  "accuracy": "91.39%"
}
```

## 📝 Örnekler

### Python

```python
import requests

url = "https://YOUR-SPACE-URL.hf.space/predict_pet"

with open("dog.jpg", "rb") as f:
    files = {"image": f}
    response = requests.post(url, files=files)
    
result = response.json()
print(f"Breed: {result['breed']}")
print(f"Confidence: {result['confidence']:.2%}")
```

### cURL

```bash
curl -X POST https://YOUR-SPACE-URL.hf.space/predict_pet \
  -F "image=@dog.jpg"
```

### JavaScript (Fetch)

```javascript
const formData = new FormData();
formData.append('image', fileInput.files[0]);

const response = await fetch('https://YOUR-SPACE-URL.hf.space/predict_pet', {
  method: 'POST',
  body: formData
});

const result = await response.json();
console.log(result.breed, result.confidence);
```

## ℹ️ Model Bilgisi

- **Model:** [Pavarissy/ConvNextV2-large-DogBreed](https://huggingface.co/Pavarissy/ConvNextV2-large-DogBreed)
- **Accuracy:** 91.39% (validation set)
- **Architecture:** ConvNextV2-large-22k-224
- **Training:** 50 epochs, Stanford Dogs Dataset
- **Classes:** 120 dog breeds

## 🔧 Performans

- **İlk istek:** 10-15 saniye (model yükleme)
- **Sonraki istekler:** 2-4 saniye
- **Hardware:** CPU basic (HF Spaces free tier)

## 📄 License

MIT