app.py

import gradio as gr
import torch
from huggingface_hub import hf_hub_download
import sys
import os
from PIL import Image
import torchvision.transforms as transforms


photos_folder = "Photos"

# Download model and config
repo_id = "Kiwinicki/sat2map-generator"
generator_path = hf_hub_download(repo_id=repo_id, filename="generator.pth")
model_path = hf_hub_download(repo_id=repo_id, filename="model.py")

# Add path to model
sys.path.append(os.path.dirname(model_path))
from model import Generator, GeneratorConfig

# Initialize model
cfg = GeneratorConfig()
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
generator = Generator(cfg).to(device)
generator.load_state_dict(torch.load(generator_path, map_location=device))
generator.eval()

# Transformations
transform = transforms.Compose([
    transforms.Resize((256, 256)),
    transforms.ToTensor(),
    transforms.Normalize(mean=[0.5, 0.5, 0.5], std=[0.5, 0.5, 0.5])
])


def process_image(image):
    if image is None:
        return None
        
    # Convert to tensor
    image_tensor = transform(image).unsqueeze(0).to(device)
    
    # Inference
    with torch.no_grad():
        output_tensor = generator(image_tensor)
    
    # Prepare output
    output_image = output_tensor.squeeze(0).cpu()
    output_image = output_image * 0.5 + 0.5  # Denormalization
    output_image = transforms.ToPILImage()(output_image)
    return output_image

def load_images_from_folder(folder):
    images = []
    if not os.path.exists(folder):
        os.makedirs(folder)
        return images
        
    for filename in os.listdir(folder):
        if filename.lower().endswith(('.png')):
            img_path = os.path.join(folder, filename)
            try:
                img = Image.open(img_path)
                images.append((img, filename))
            except Exception as e:
                print(f"Error loading {filename}: {e}")
    return images

def app():
    images = load_images_from_folder(photos_folder)
    gallery_images = [img[0] for img in images] if images else []

    with gr.Blocks() as demo:
        with gr.Row():
            with gr.Column():
                input_image = gr.Image(label="Input Image", type="pil")
                clear_button = gr.Button("Clear")

            with gr.Column():
                gallery = gr.Gallery(
                    label="Image Gallery",
                    value=gallery_images,
                    columns=3,
                    rows=2,
                    height="auto"
                )

            with gr.Column():
                output_image = gr.Image(label="Result Image", type="pil")
        
        # Handle gallery selection
        def on_select(evt: gr.SelectData):
            if 0 <= evt.index < len(images):
                return images[evt.index][0]
            return None
            
        gallery.select(
            fn=on_select,
            outputs=input_image
        )
        
        # Process image when input changes
        input_image.change(
            fn=process_image,
            inputs=input_image,
            outputs=output_image
        )
        
        # Clear button functionality
        clear_button.click(
            fn=lambda: None,
            outputs=input_image
        )

    demo.launch()

if __name__ == "__main__":
    app()