mask_app.py

import os
import gradio as gr
from gradio_imageslider import ImageSlider
from loadimg import load_img
#import spaces
from transformers import AutoModelForImageSegmentation
import torch
from torchvision import transforms
import numpy as np
from PIL import Image

# 检查 CUDA 是否可用
if torch.cuda.is_available():
    device = "cuda"
else:
    device = "cpu"

torch.set_float32_matmul_precision(["high", "highest"][0])

birefnet = AutoModelForImageSegmentation.from_pretrained(
    "briaai/RMBG-2.0", trust_remote_code=True
)
birefnet.to(device)
transform_image = transforms.Compose(
    [
        transforms.Resize((1024, 1024)),
        transforms.ToTensor(),
        transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225]),
    ]
)

output_folder = 'output_images'
if not os.path.exists(output_folder):
    os.makedirs(output_folder)

# 定义颜色列表，每个颜色对应一个 mask
colors = [
    '#000000',  # 背景色
    '#2692F3',  # 蓝色
    '#F89E12',  # 橙色
    '#16C232',  # 绿色
    '#F92F6C',  # 粉色
    '#AC6AEB',  # 紫色
]

# 将颜色转换为 RGB 值
palette = np.array([
    tuple(int(s[i + 1:i + 3], 16) for i in (0, 2, 4))
    for s in colors[1:]  # 跳过背景色
])  # (N, 3)

def fn(image, mask_color):
    im = load_img(image, output_type="pil")
    im = im.convert("RGB")
    origin = im.copy()
    image, mask = process(im, mask_color)    
    image_path = os.path.join(output_folder, "no_bg_image.png")
    mask_path = os.path.join(output_folder, "mask_image.png")
    image.save(image_path)
    mask.save(mask_path)
    return (image, origin), image_path, mask

#@spaces.GPU
def process(image, mask_color):
    image_size = image.size
    input_images = transform_image(image).unsqueeze(0).to(device)
    # Prediction
    with torch.no_grad():
        preds = birefnet(input_images)[-1].sigmoid().cpu()
    pred = preds[0].squeeze()
    pred_pil = transforms.ToPILImage()(pred)
    mask = pred_pil.resize(image_size)
    
    # 创建一个新的透明背景图像
    transparent_image = Image.new("RGBA", image_size, (0, 0, 0, 0))
    transparent_image.paste(image, (0, 0), mask)
    
    # 创建一个带有颜色的 mask 图像
    mask_color_rgb = tuple(int(mask_color[i + 1:i + 3], 16) for i in (0, 2, 4))
    colored_mask = Image.new("RGBA", image_size, mask_color_rgb + (255,))
    colored_mask.putalpha(mask)
    
    return transparent_image, colored_mask

# 示例数据
example_image = "giraffe.jpg"  # 确保该文件存在于当前目录
example_url = "http://farm9.staticflickr.com/8488/8228323072_76eeddfea3_z.jpg"

# 定义 Gradio 组件
with gr.Blocks() as demo:
    gr.Markdown("# 🖼️ RMBG-2.0 for Background Removal")
    with gr.Row():
        # 左侧列：输入
        with gr.Column():
            gr.Markdown("## Input")
            image_input = gr.Image(label="Upload an image")
            text_input = gr.Textbox(label="Paste an image URL")
            color_input = gr.Dropdown(label="Mask Color", choices=colors[1:], value=colors[1])
            run_button = gr.Button("Run")

        # 右侧列：输出
        with gr.Column():
            gr.Markdown("## Output")
            slider_output = ImageSlider(label="RMBG-2.0", type="pil")
            file_output = gr.File(label="Output PNG File")
            mask_output = gr.Image(label="Mask Image")

    # 示例数据
    gr.Examples(
        examples=[[example_image, colors[1]], [example_url, colors[1]]],
        inputs=[image_input, color_input],
        outputs=[slider_output, file_output, mask_output],  # 添加 outputs 参数
        fn=fn,
        cache_examples=True
    )

    # 绑定事件
    run_button.click(
        fn=fn,
        inputs=[image_input, color_input],
        outputs=[slider_output, file_output, mask_output]
    )

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