app.py

"""
Waldo and Wilma Detector - Flask Web App for Hugging Face Spaces
Detects Waldo and Wilma in uploaded images using Grounding DINO
"""

from flask import Flask, render_template, request, jsonify, send_from_directory
import torch
from PIL import Image, ImageDraw, ImageFont
from transformers import AutoProcessor, AutoModelForZeroShotObjectDetection
import numpy as np
import os
import io
import base64
from werkzeug.utils import secure_filename
import uuid

app = Flask(__name__)
app.config['MAX_CONTENT_LENGTH'] = 16 * 1024 * 1024  # 16MB max file size
app.config['UPLOAD_FOLDER'] = 'uploads'
app.config['ALLOWED_EXTENSIONS'] = {'png', 'jpg', 'jpeg', 'gif', 'bmp'}

# Create upload folder if it doesn't exist
os.makedirs(app.config['UPLOAD_FOLDER'], exist_ok=True)

class WaldoWilmaDetector:
    """Detector for Waldo and Wilma using Grounding DINO."""
    
    def __init__(self):
        self.device = "cuda" if torch.cuda.is_available() else "cpu"
        print(f"Loading model on {self.device}...")
        
        model_id = "IDEA-Research/grounding-dino-base"
        self.processor = AutoProcessor.from_pretrained(model_id)
        self.model = AutoModelForZeroShotObjectDetection.from_pretrained(model_id).to(self.device)
        
        print("Model loaded successfully!")
        
    def detect(self, image, box_threshold=0.35, text_threshold=0.25):
        """
        Detect Waldo and Wilma in an image.
        
        Args:
            image: PIL Image
            box_threshold: Confidence threshold for bounding boxes
            text_threshold: Confidence threshold for text matching
            
        Returns:
            tuple: (annotated_image, detection_info)
        """
        # Ensure image is RGB
        if image.mode != "RGB":
            image = image.convert("RGB")
        
        # Text prompt optimized for Waldo and Wilma detection
        text_prompt = (
            "a person wearing red and white horizontal striped shirt with glasses and beanie. "
            "a person wearing red and white striped shirt. "
            "a character with red white stripes."
        )
        
        # Process image
        inputs = self.processor(images=image, text=text_prompt, return_tensors="pt").to(self.device)
        
        # Run detection
        with torch.no_grad():
            outputs = self.model(**inputs)
        
        # Post-process results
        results = self.processor.post_process_grounded_object_detection(
            outputs,
            inputs.input_ids,
            text_threshold=text_threshold,
            target_sizes=[image.size[::-1]]
        )
        
        # Visualize results
        annotated_image = self._draw_boxes(image.copy(), results[0])
        
        # Get detection info
        num_detections = len(results[0]["boxes"])
        boxes = results[0]["boxes"].tolist()
        scores = results[0]["scores"].tolist()
        
        detection_info = {
            'num_detections': num_detections,
            'boxes': boxes,
            'scores': scores,
            'summary': f"Found {num_detections} character(s) wearing red and white stripes!"
        }
        
        return annotated_image, detection_info
    
    def _draw_boxes(self, image, result):
        """Draw bounding boxes on image."""
        draw = ImageDraw.Draw(image)
        
        # Try to load a font, fallback to default if not available
        try:
            font = ImageFont.truetype("arial.ttf", 20)
        except:
            font = ImageFont.load_default()
        
        boxes = result["boxes"]
        scores = result["scores"]
        labels = result["labels"]
        
        # Define colors for different detections
        colors = ["red", "blue", "green", "yellow", "purple"]
        
        for idx, (box, score, label) in enumerate(zip(boxes, scores, labels)):
            # Convert box coordinates to integers
            box = [int(i) for i in box.tolist()]
            color = colors[idx % len(colors)]
            
            # Draw rectangle
            draw.rectangle(box, outline=color, width=4)
            
            # Draw label with background
            text_label = f"Waldo/Wilma: {score:.2%}"
            
            # Get text bounding box for background
            bbox = draw.textbbox((box[0], box[1] - 25), text_label, font=font)
            draw.rectangle(bbox, fill=color)
            draw.text((box[0], box[1] - 25), text_label, fill="white", font=font)
        
        return image

# Initialize detector
print("Initializing Waldo & Wilma Detector...")
detector = WaldoWilmaDetector()
print("Detector ready!")

def allowed_file(filename):
    """Check if file extension is allowed."""
    return '.' in filename and \
           filename.rsplit('.', 1)[1].lower() in app.config['ALLOWED_EXTENSIONS']

def image_to_base64(image):
    """Convert PIL Image to base64 string."""
    buffered = io.BytesIO()
    image.save(buffered, format="PNG")
    img_str = base64.b64encode(buffered.getvalue()).decode()
    return f"data:image/png;base64,{img_str}"

@app.route('/')
def index():
    """Render the main page."""
    return render_template('index.html')

@app.route('/detect', methods=['POST'])
def detect():
    """Handle image upload and detection."""
    try:
        # Check if image was uploaded
        if 'image' not in request.files:
            return jsonify({'error': 'No image uploaded'}), 400
        
        file = request.files['image']
        
        if file.filename == '':
            return jsonify({'error': 'No image selected'}), 400
        
        if not allowed_file(file.filename):
            return jsonify({'error': 'Invalid file type. Please upload an image.'}), 400
        
        # Get threshold parameters
        box_threshold = float(request.form.get('box_threshold', 0.35))
        text_threshold = float(request.form.get('text_threshold', 0.25))
        
        # Read and process image
        image = Image.open(file.stream)
        
        # Run detection
        annotated_image, detection_info = detector.detect(
            image,
            box_threshold=box_threshold,
            text_threshold=text_threshold
        )
        
        # Convert annotated image to base64
        result_image_base64 = image_to_base64(annotated_image)
        
        return jsonify({
            'success': True,
            'image': result_image_base64,
            'detections': detection_info
        })
        
    except Exception as e:
        print(f"Error during detection: {str(e)}")
        return jsonify({'error': f'Detection failed: {str(e)}'}), 500

@app.route('/health')
def health():
    """Health check endpoint."""
    return jsonify({'status': 'healthy', 'model_loaded': True})

if __name__ == '__main__':
    # For local development
    app.run(host='0.0.0.0', port=7860, debug=False)