import os import warnings from pathlib import Path import gradio as gr from gradio_imageslider import ImageSlider import numpy as np import torch import fastai from deoldify import device from deoldify.device_id import DeviceId from deoldify.visualize import * from huggingface_hub import HfApi, HfFolder os.system("pip freeze") from collections.abc import Sized # Import Sized from collections.abc # Suppress warnings warnings.filterwarnings("ignore", category=UserWarning, message=".*?Your .*? set is empty.*?") repo_id = "afondiel/image-colorizer-deoldify" repo_type = "space" # run on CPU device.set(device=DeviceId.CPU) # Initialize Hugging Face API api = HfApi() # Download the snapshot from the space repository snapshot_folder = api.snapshot_download(repo_id=repo_id, repo_type=repo_type) # if GPU available device.set(device=DeviceId.GPU0) # Load the pre-trained model _colorizer = get_image_colorizer(root_folder=Path(snapshot_folder), artistic=True) def colorizer_fn(input_img, render_factor): """ Colorize grayscale images/photos - @param input_img old (grayscale) image - @param render_factor render_factor """ if input_img is not None and input_img != '': output_img = _colorizer.get_transformed_image( path=input_img, render_factor=render_factor, watermarked=False, post_process=True, ) else: print('Provide an image and try again.') return (input_img, output_img) # Return a tuple of old and color Image to be plotted with ImageSlider() title = "AI Image Colorizer" description = "Colorize old images with AI" examples = [["./demo.jpg"],] demo = gr.Interface( fn=colorizer_fn, inputs=[gr.Image(type="filepath", label="Old image"), gr.Slider(minimum=0, maximum=40, step=1, label="Render Factor", value=10)], outputs=ImageSlider(type="pil", label="Old vs Colored image"), examples=examples, title=title, description=description, ) # Launch the demo if __name__ == "__main__": demo.launch()