app.py

import io
import requests
import numpy as np
import torch
import os
from PIL import Image
from typing import List, Optional
from functools import reduce
from argparse import ArgumentParser

import gradio as gr

from transformers import DetrFeatureExtractor, DetrForSegmentation, DetrConfig
from transformers.models.detr.feature_extraction_detr import rgb_to_id

from diffusers import StableDiffusionInpaintPipeline, DPMSolverMultistepScheduler

parser = ArgumentParser()
parser.add_argument('--disable-cuda', action='store_true')
parser.add_argument('--attention-slicing', action='store_true')
args = parser.parse_args()

auth_token = os.environ.get("READ_TOKEN")
try_cuda = not args.disable_cuda

torch.inference_mode()
torch.no_grad()

# Device helper
def get_device(try_cuda=True):
    return torch.device('cuda' if try_cuda and torch.cuda.is_available() else 'cpu')
    
device = get_device(try_cuda=try_cuda)

# Load segmentation models
def load_segmentation_models(model_name: str = 'facebook/detr-resnet-50-panoptic'):
    feature_extractor = DetrFeatureExtractor.from_pretrained(model_name)
    model = DetrForSegmentation.from_pretrained(model_name)
    cfg = DetrConfig.from_pretrained(model_name)

    return feature_extractor, model, cfg

# Load diffusion pipeline
def load_diffusion_pipeline(model_name: str = 'stabilityai/stable-diffusion-2-inpainting'):
    return StableDiffusionInpaintPipeline.from_pretrained(
        model_name,
        revision='fp16',
        torch_dtype=torch.float16 if try_cuda and torch.cuda.is_available() else torch.float32,
        use_auth_token=auth_token
    )

def min_pool(x: torch.Tensor, kernel_size: int):
    pad_size = (kernel_size - 1) // 2
    return -torch.nn.functional.max_pool2d(-x, kernel_size, (1, 1), padding=pad_size) 

def max_pool(x: torch.Tensor, kernel_size: int):
    pad_size = (kernel_size - 1) // 2
    return torch.nn.functional.max_pool2d(x, kernel_size, (1, 1), padding=pad_size) 

# Apply min-max pooling to clean up mask
def clean_mask(mask, max_kernel: int = 23, min_kernel: int = 5):
    mask = torch.Tensor(mask[None, None]).float().to(device)
    mask = min_pool(mask, min_kernel)
    mask = max_pool(mask, max_kernel)
    mask = mask.bool().squeeze().cpu().numpy()
    return mask


feature_extractor, segmentation_model, segmentation_cfg = load_segmentation_models()
pipe = load_diffusion_pipeline()
pipe.scheduler = DPMSolverMultistepScheduler.from_config(pipe.scheduler.config)

segmentation_model = segmentation_model.to(device)
pipe = pipe.to(device)
if args.attention_slicing:
    pipe.enable_attention_slicing()

# Callback function that runs segmentation and updates CheckboxGroup
def fn_segmentation(image, max_kernel, min_kernel):
    inputs = feature_extractor(images=image, return_tensors="pt").to(device)
    outputs = segmentation_model(**inputs)

    processed_sizes = torch.as_tensor(inputs["pixel_values"].shape[-2:]).unsqueeze(0)
    result = feature_extractor.post_process_panoptic(outputs, processed_sizes)[0]

    panoptic_seg = Image.open(io.BytesIO(result["png_string"])).resize((image.width, image.height))
    panoptic_seg = np.array(panoptic_seg, dtype=np.uint8)

    panoptic_seg_id = rgb_to_id(panoptic_seg)

    raw_masks = []
    for s in result['segments_info']:
        m = panoptic_seg_id == s['id']
        raw_masks.append(m.astype(np.uint8) * 255)
    
    checkbox_choices = [f"{s['id']}:{segmentation_cfg.id2label[s['category_id']]}" for s in result['segments_info']]
    
    checkbox_group = gr.CheckboxGroup.update(
        choices=checkbox_choices
    )

    return raw_masks, checkbox_group, gr.Image.update(value=np.zeros((image.height, image.width))), gr.Image.update(value=image)

# Callback function that updates the displayed mask based on selected checkboxes
def fn_update_mask(
        image: Image,
        masks: List[np.array], 
        masks_enabled: List[int], 
        max_kernel: int,
        min_kernel: int,
        invert_mask: bool
    ):
    masks_enabled = [int(m.split(':')[0]) for m in masks_enabled]
    combined_mask = reduce(lambda x, y: x | y, [masks[i] for i in masks_enabled], np.zeros_like(masks[0], dtype=bool))

    if invert_mask:
        combined_mask = ~combined_mask

    combined_mask = clean_mask(combined_mask, max_kernel, min_kernel)

    masked_image = np.array(image).copy()
    masked_image[combined_mask] = 0.0

    return combined_mask.astype(np.uint8) * 255, Image.fromarray(masked_image)

# Callback function that runs diffusion given the current image, mask and prompt.
def fn_diffusion(
        prompt: str, 
        masked_image: Image, 
        mask: Image, 
        num_diffusion_steps: int,
        guidance_scale: float,
        negative_prompt: Optional[str] = None,
    ):
    if len(negative_prompt) == 0:
        negative_prompt = None

    # Resize image to a more stable diffusion friendly format.
    # TODO: remove magic number
    STABLE_DIFFUSION_SMALL_EDGE = 512

    w, h = masked_image.size
    is_width_larger = w > h
    resize_ratio = STABLE_DIFFUSION_SMALL_EDGE / (h if is_width_larger else w)

    new_width = int(w * resize_ratio) if is_width_larger else STABLE_DIFFUSION_SMALL_EDGE
    new_height = STABLE_DIFFUSION_SMALL_EDGE if is_width_larger else int(h * resize_ratio)

    new_width += 8 - (new_width % 8) if is_width_larger else 0
    new_height += 0 if is_width_larger else 8 - (new_height % 8)

    mask = Image.fromarray(mask).convert("RGB").resize((new_width, new_height))
    masked_image = masked_image.convert("RGB").resize((new_width, new_height))

    # Run diffusion
    inpainted_image = pipe(
        height=new_height, 
        width=new_width, 
        prompt=prompt,
        image=masked_image, 
        mask_image=mask,
        num_inference_steps=num_diffusion_steps,
        guidance_scale=guidance_scale,
        negative_prompt=negative_prompt
    ).images[0]

    # Resize back to the original size
    inpainted_image = inpainted_image.resize((w, h))

    return inpainted_image

demo = gr.Blocks(css=open('app.css').read())

with demo:
    gr.HTML(open('app_header.html').read())

    if not try_cuda or not torch.cuda.is_available():
        gr.HTML('<div class="alert alert-warning" role="alert" style="color:red"><b>Warning: GPU not available! Diffusion will be slow.</b></div>')

    # Input image control
    input_image = gr.Image(value="example.png", type='pil', label="Input Image")
    # Combined mask controls
    bt_masks = gr.Button("Compute Masks")
    with gr.Row():
        mask_image = gr.Image(type='numpy', label="Diffusion Mask")
        masked_image = gr.Image(type='pil', label="Masked Image")
    mask_storage = gr.State()

    # Mask editing controls
    with gr.Row():
        max_slider = gr.Slider(minimum=1, maximum=99, value=23, step=2, label="Mask Overflow")
        min_slider = gr.Slider(minimum=1, maximum=99, value=5, step=2, label="Mask Denoising")
        
    with gr.Row():  
        invert_mask = gr.Checkbox(label="Invert Mask")
        mask_checkboxes = gr.CheckboxGroup(interactive=True, label="Mask Selection")

    # Diffusion controls and output
    with gr.Row():
        with gr.Column():
            prompt = gr.Textbox("An angry dog floating in outer deep space. Twinkling stars in the background. High definition.", label="Prompt")
            negative_prompt = gr.Textbox(label="Negative Prompt")
        with gr.Column():
            steps_slider = gr.Slider(minimum=1, maximum=100, value=50, label="Inference Steps")
            guidance_slider = gr.Slider(minimum=0.0, maximum=50.0, value=7.5, step=0.1, label="Guidance Scale")
            bt_diffusion = gr.Button("Run Diffusion")

        inpainted_image = gr.Image(type='pil', label="Inpainted Image")

    # TODO: saw a better way of handling many inputs online..
    # forgot where though
    update_mask_inputs = [input_image, mask_storage, mask_checkboxes, max_slider, min_slider, invert_mask]
    update_mask_outputs = [mask_image, masked_image]

    # Clear checkbox group on input image change
    input_image.change(lambda: gr.CheckboxGroup.update(choices=[], value=[]), outputs=mask_checkboxes)
    input_image.change(lambda: gr.Checkbox.update(value=False), outputs=invert_mask)

    # Segmentation button callback
    bt_masks.click(fn_segmentation, inputs=[input_image, max_slider, min_slider], outputs=[mask_storage, mask_checkboxes, mask_image, masked_image])

    # Update mask callbacks
    max_slider.change(fn_update_mask, inputs=update_mask_inputs, outputs=update_mask_outputs, show_progress=False)
    min_slider.change(fn_update_mask, inputs=update_mask_inputs, outputs=update_mask_outputs, show_progress=False)
    mask_checkboxes.change(fn_update_mask, inputs=update_mask_inputs, outputs=update_mask_outputs, show_progress=False)
    invert_mask.change(fn_update_mask, inputs=update_mask_inputs, outputs=update_mask_outputs, show_progress=False)

    # Diffusion button callback
    bt_diffusion.click(fn_diffusion, inputs=[
        prompt, 
        masked_image, 
        mask_image, 
        steps_slider, 
        guidance_slider, 
        negative_prompt
    ], outputs=inpainted_image)
    gr.HTML(open('app_license.html').read())

demo.launch()