app.py

import gradio as gr
from gradio_imageslider import ImageSlider
from loadimg import load_img
from transformers import AutoModelForImageSegmentation
import torch
from torchvision import transforms
import os
import zipfile
from PIL import Image

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

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

try:
    birefnet = AutoModelForImageSegmentation.from_pretrained(
        "ZhengPeng7/BiRefNet", trust_remote_code=True
    )
    birefnet.to("cpu")
except Exception as e:
    print(f"Error loading model: {e}")
    raise

transform_image = transforms.Compose(
    [
        transforms.Resize((1024, 1024)),
        transforms.ToTensor(),
        transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225]),
    ]
)

def process_single_image(image, output_type="mask"):
    if image is None:
        return [None, None], None
    
    im = load_img(image, output_type="pil")
    if im is None:
        return [None, None], None

    im = im.convert("RGB")
    image_size = im.size
    origin = im.copy()
    input_images = transform_image(im).unsqueeze(0).to("cpu")
    
    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)
    
    processed_im = im.copy()
    processed_im.putalpha(mask)
    output_file_path = os.path.join(output_folder, "output_image_i2i.png")
    processed_im.save(output_file_path)

    if output_type == "origin":
        return [processed_im, origin], output_file_path
    else:
        return [processed_im, mask], output_file_path

def process_image_from_url(url, output_type="mask"):
    if url is None or url.strip() == "":
        return [None, None], None
    
    try:
        im = load_img(url, output_type="pil")
        if im is None:
            return [None, None], None

        im = im.convert("RGB")
        image_size = im.size
        origin = im.copy()
        input_images = transform_image(im).unsqueeze(0).to("cpu")
        
        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)
        
        processed_im = im.copy()
        processed_im.putalpha(mask)
        output_file_path = os.path.join(output_folder, "output_image_url.png")
        processed_im.save(output_file_path)

        if output_type == "origin":
            return [processed_im, origin], output_file_path
        else:
            return [processed_im, mask], output_file_path
    except Exception as e:
        return [None, None], str(e)

def process_batch_images(images):
    output_paths = []
    if not images:
        return [], None
    
    for idx, image_path in enumerate(images):
        im = load_img(image_path, output_type="pil")
        if im is None:
            continue
        
        im = im.convert("RGB")
        image_size = im.size
        input_images = transform_image(im).unsqueeze(0).to("cpu")
        
        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)
        
        im.putalpha(mask)
        output_file_path = os.path.join(output_folder, f"output_image_batch_{idx + 1}.png")
        im.save(output_file_path)
        output_paths.append(output_file_path)

    zip_file_path = os.path.join(output_folder, "processed_images.zip")
    with zipfile.ZipFile(zip_file_path, 'w') as zipf:
        for file in output_paths:
            zipf.write(file, os.path.basename(file))

    return output_paths, zip_file_path

image = gr.Image(label="Upload an image")
text = gr.Textbox(label="Paste an image URL")
batch_image = gr.File(label="Upload multiple images", type="filepath", file_count="multiple")

slider1 = ImageSlider(label="Processed Image", type="pil")
slider2 = ImageSlider(label="Processed Image from URL", type="pil")

tab1 = gr.Interface(
    fn=process_single_image, 
    inputs=[image, gr.Radio(choices=["mask", "origin"], value="mask", label="Select Output Type")],
    outputs=[slider1, gr.File(label="PNG Output")],
    examples=[["chameleon.jpg"]],
    api_name="image"
)

tab2 = gr.Interface(
    fn=process_image_from_url, 
    inputs=[text, gr.Radio(choices=["mask", "origin"], value="mask", label="Select Output Type")],
    outputs=[slider2, gr.File(label="PNG Output")],
    examples=[["https://hips.hearstapps.com/hmg-prod/images/gettyimages-1229892983-square.jpg"]], 
    api_name="text"
)

tab3 = gr.Interface(
    fn=process_batch_images, 
    inputs=batch_image, 
    outputs=[gr.Gallery(label="Processed Images"), gr.File(label="Download Processed Files")], 
    api_name="batch"
)

demo = gr.TabbedInterface(
    [tab1, tab2, tab3], 
    ["image", "text", "batch"], 
    title="Multi Birefnet for Background Removal"
)

if __name__ == "__main__":
    demo.launch()