app.py

import json
import os

import numpy as np
import torch
import gradio as gr

from config import PipelineConfig
from src.pipeline import FashionPipeline, PipelineOutput


config = PipelineConfig()
fashion_pipeline = FashionPipeline(config, device=torch.device('cuda'))


def process(
    input_image: np.ndarray,
    prompt: str,
    negative_prompt: str,
    generate_from_mask: bool,
    num_inference_steps: int,
    guidance_scale: float,
    conditioning_scale: float,
    target_image_size: int,
    max_image_size: int,
    seed: int,
):
    output: PipelineOutput = fashion_pipeline(
        control_image=input_image,
        prompt=prompt,
        negative_prompt=negative_prompt,
        generate_from_mask=generate_from_mask,
        num_inference_steps=num_inference_steps,
        guidance_scale=float(guidance_scale),
        conditioning_scale=float(conditioning_scale),
        target_image_size=target_image_size,
        max_image_size=max_image_size,
        seed=seed,
    )
    return [
        output.generated_image,
        output.control_mask,
    ]


def read_content(file_path: str) -> str:
    """Read the content of target file."""
    with open(file_path, 'r', encoding='utf-8') as f:
        content = f.read()

    return content


image_dir = 'examples/images'
image_list = [os.path.join(image_dir, file) for file in os.listdir(image_dir)]
with open('examples/prompts.json', 'r') as f:
    prompts_list = json.load(f).values()
examples = [[image, prompt[0], prompt[1]] for image, prompt in zip(image_list, prompts_list)]


block = gr.Blocks().queue()
with block:
    with gr.Row():
        gr.HTML(read_content('header.html'))
    with gr.Row():
        with gr.Column():
            input_image = gr.Image(type='numpy')
            prompt = gr.Textbox(label='Prompt')
            negative_prompt = gr.Textbox(label='Negative Prompt')
            with gr.Row():
                generate_from_mask = gr.Checkbox(label='Input image is already a control mask', value=False)
            run_button = gr.Button(value='Run')
            with gr.Accordion('Advanced options', open=False):
                target_image_size = gr.Slider(
                    label='Image target size:',
                    minimum=512,
                    maximum=2048,
                    value=768,
                    step=64,
                )
                max_image_size = gr.Slider(
                    label='Image max size:',
                    minimum=512,
                    maximum=2048,
                    value=1024,
                    step=64,
                )
                num_inference_steps = gr.Slider(label='Number of steps', minimum=1, maximum=100, value=20, step=1)
                guidance_scale = gr.Slider(label='Guidance scale', minimum=0.1, maximum=30.0, value=9.0, step=0.1)
                conditioning_scale = gr.Slider(label='Conditioning scale', minimum=0.0, maximum=5.0, value=1.0, step=0.1)
                seed = gr.Slider(label='Seed', minimum=0, maximum=config.max_seed, step=1, value=0)

            gr.Examples(examples=examples, inputs=[input_image, prompt, negative_prompt], label='Examples - Input Images', examples_per_page=12)

            gr.HTML(
                """
                    <div class="footer">
                    
                        <p> This repo based on Unet from <a style="text-decoration: underline;" href="https://huggingface.co/spaces/wildoctopus/cloth-segmentation">cloth-segmentation</a>
        It's uses pre-trained U2NET to extract Upper body(red), Lower body(green), Full body(blue) masks, and then
        run StableDiffusionXLControlNetPipeline with trained controlnet_baseline to generate image conditioned on this masks.
                        </p>
                """)

        with gr.Column():
            generated_output = gr.Image(label='Generated', type='numpy', elem_id='generated')
            mask_output = gr.Image(label='Mask', type='numpy', elem_id='mask')

    ips = [input_image, prompt, negative_prompt, generate_from_mask, num_inference_steps, guidance_scale, conditioning_scale, target_image_size, max_image_size, seed]
    run_button.click(fn=process, inputs=ips, outputs=[generated_output, mask_output])


block.launch()