app.py

"""
Clean Pet Backend for Hugging Face Spaces
Using ConvNextV2-large-DogBreed model
"""

from flask import Flask, request, jsonify
from flask_cors import CORS
from PIL import Image
import io
import os

# Set cache directories FIRST
os.environ['HF_HOME'] = '/tmp/.cache'
os.environ['TRANSFORMERS_CACHE'] = '/tmp/.cache'
os.environ['TORCH_HOME'] = '/tmp/.cache'

app = Flask(__name__)
CORS(app)

# Global model variables
model = None
processor = None

def load_model():
    """Load model on first request"""
    global model, processor
    
    if model is not None:
        return
    
    print("🔄 Loading ConvNextV2-large-DogBreed model...")
    
    from transformers import AutoImageProcessor, AutoModelForImageClassification
    import torch
    
    model_name = "Pavarissy/ConvNextV2-large-DogBreed"
    
    processor = AutoImageProcessor.from_pretrained(model_name)
    model = AutoModelForImageClassification.from_pretrained(model_name)
    
    device = "cuda" if torch.cuda.is_available() else "cpu"
    model = model.to(device)
    model.eval()
    
    print(f"✅ Model loaded on {device}")

@app.route('/', methods=['GET'])
def health():
    return jsonify({
        'status': 'healthy',
        'model': 'ConvNextV2-large-DogBreed',
        'accuracy': '91.39%'
    })

@app.route('/predict_pet', methods=['POST'])
def predict_pet():
    try:
        # Load model if needed
        load_model()
        
        # Get image
        if 'image' not in request.files:
            return jsonify({'error': 'No image provided'}), 400
        
        file = request.files['image']
        image = Image.open(io.BytesIO(file.read())).convert('RGB')
        
        # Process and predict
        import torch
        inputs = processor(image, return_tensors="pt")
        
        device = next(model.parameters()).device
        inputs = {k: v.to(device) for k, v in inputs.items()}
        
        with torch.no_grad():
            outputs = model(**inputs)
            logits = outputs.logits
        
        # Get top prediction
        probs = torch.nn.functional.softmax(logits, dim=-1)[0]
        top_prob, top_idx = torch.max(probs, dim=0)
        
        predicted_breed = model.config.id2label[top_idx.item()]
        confidence = float(top_prob.item())
        
        print(f"✅ Predicted: {predicted_breed} ({confidence:.2%})")
        
        # Return in format expected by frontend
        return jsonify({
            'predicted_label': predicted_breed,
            'breed': predicted_breed,
            'confidence': confidence,
            'detection': {'box': {'x': 0, 'y': 0, 'width': 0, 'height': 0}},
            'gender': 'Unknown'
        })
        
    except Exception as e:
        print(f"❌ Error: {str(e)}")
        import traceback
        traceback.print_exc()
        return jsonify({'error': str(e)}), 500

if __name__ == '__main__':
    port = int(os.environ.get('PORT', 7860))
    print(f"🚀 Starting server on port {port}")
    app.run(host='0.0.0.0', port=port, debug=False)