app.py

import torch
import torch.nn as nn
from torchvision import transforms
from torchvision.models import convnext_tiny
from ultralytics import YOLO
import numpy as np
import cv2
import gradio as gr
from PIL import Image, ImageDraw
from fast_alpr import ALPR

# ------------------ Constants and Models ------------------
class_names = [
    'beige', 'black', 'blue', 'brown', 'gold',
    'green', 'grey', 'orange', 'pink', 'purple',
    'red', 'silver', 'tan', 'white', 'yellow'
]

DETECTOR_MODEL = "yolo-v9-s-608-license-plate-end2end"
OCR_MODEL = "global-plates-mobile-vit-v2-model"

device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
model = convnext_tiny(pretrained=False)
model.classifier[2] = nn.Linear(768, len(class_names))
model.load_state_dict(torch.load("convnext_best_model.pth", map_location=device))
model = model.to(device)
model.eval()

transform = transforms.Compose([
    transforms.Resize((512, 512)),
    transforms.ToTensor(),
    transforms.Normalize([0.485, 0.456, 0.406],
                         [0.229, 0.224, 0.225])
])

yolo_model = YOLO("yolo11x.pt")

# ------------------ Unified Inference Function ------------------
def alpr_color_inference(image):
    if image is None:
        return None, None, None, "Please upload an image to continue."

    img = image.convert("RGB")
    img_array = np.array(img)
    alpr = ALPR(detector_model=DETECTOR_MODEL, ocr_model=OCR_MODEL)
    results = alpr.predict(img_array)

    annotated_img = Image.fromarray(img_array.copy())
    draw = ImageDraw.Draw(annotated_img)

    plate_texts = []
    for result in results:
        detection = getattr(result, 'detection', None)
        ocr = getattr(result, 'ocr', None)
        if detection is not None:
            bbox_obj = getattr(detection, 'bounding_box', None)
            if bbox_obj is not None:
                bbox = [int(bbox_obj.x1), int(bbox_obj.y1), int(bbox_obj.x2), int(bbox_obj.y2)]
                draw.rectangle(bbox, outline="red", width=3)
                if ocr is not None:
                    text = getattr(ocr, 'text', '')
                    plate_texts.append(text)
                    draw.text((bbox[0], max(bbox[1] - 10, 0)), text, fill="red")

    # Color Detection
    img_cv2 = cv2.cvtColor(np.array(img), cv2.COLOR_RGB2BGR)
    yolo_results = yolo_model(img_cv2)
    boxes = yolo_results[0].boxes
    vehicle_class_ids = {2, 3, 5, 7}  # car, motorcycle, bus, truck
    vehicle_boxes = [box for box in boxes if int(box.cls.item()) in vehicle_class_ids]

    if not vehicle_boxes:
        color_text = "No vehicle detected"
        cropped_img = img
        vehicle_img = img
    else:
        largest_vehicle = max(vehicle_boxes, key=lambda box: (box.xyxy[0][2] - box.xyxy[0][0]) * (box.xyxy[0][3] - box.xyxy[0][1]))
        x1, y1, x2, y2 = map(int, largest_vehicle.xyxy[0].tolist())
        cropped_img = img.crop((x1, y1, x2, y2))
        input_tensor = transform(cropped_img).unsqueeze(0).to(device)
        with torch.no_grad():
            output = model(input_tensor)
            probs = torch.softmax(output, dim=1)[0]
            pred_idx = torch.argmax(probs).item()
            pred_class = class_names[pred_idx]
            confidence = probs[pred_idx].item()
        vehicle_img = Image.fromarray(cv2.rectangle(np.array(img), (x1, y1), (x2, y2), (255, 0, 0), 3))
        color_text = f"{pred_class} ({confidence*100:.1f}%)"

    detection_results = (f"Detected {len(results)} license plate(s): {', '.join(plate_texts)}"
                         if results else "No license plate detected 😔.")

    return annotated_img, vehicle_img, cropped_img, f"{detection_results}\nVehicle Color: {color_text}"

# ------------------ Gradio UI ------------------
with gr.Blocks() as demo:
    gr.Markdown("# License Plate + Vehicle Color Detection")
    with gr.Row():
        with gr.Column():
            image_input = gr.Image(type="pil", label="Upload an image")
            submit_btn = gr.Button("Run Detection")
        with gr.Column():
            plate_output = gr.Image(label="License Plate Detection")
            vehicle_output = gr.Image(label="Detected Vehicle in Original")
            cropped_output = gr.Image(label="Cropped Vehicle Region")
            result_text = gr.Markdown(label="Results")

    submit_btn.click(
        alpr_color_inference,
        inputs=[image_input],
        outputs=[plate_output, vehicle_output, cropped_output, result_text]
    )

    gr.Examples(
        examples=[
            "examples/car1.jpg",
            "examples/car2.jpg",
            "examples/car3.jpg",
            "examples/car4.jpg",
        ],
        inputs=[image_input],
        label="Example Images"
    )

if __name__ == "__main__":
    demo.launch(share=True)