app.py

import io
import numpy as np
import torch
from PIL import Image, ImageFilter
from torchvision import transforms
import gradio as gr
from transformers import AutoModelForImageSegmentation, pipeline

# ----------------------------
# Global Setup and Model Loading
# ----------------------------

device = "cuda" if torch.cuda.is_available() else "cpu"

# Load the segmentation model (RMBG-2.0)
segmentation_model = AutoModelForImageSegmentation.from_pretrained(
    'briaai/RMBG-2.0',
    trust_remote_code=True
)
segmentation_model.to(device)
segmentation_model.eval()

# Transformation for segmentation (resizes to 512 for the model input)
image_size = (512, 512)
segmentation_transform = transforms.Compose([
    transforms.Resize(image_size),
    transforms.ToTensor(),
    transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
])

# Load the depth estimation pipeline (Depth-Anything)
depth_pipeline = pipeline("depth-estimation", model="depth-anything/Depth-Anything-V2-Small-hf")

# ----------------------------
# Processing Functions
# ----------------------------

def segment_and_blur_background(input_image: Image.Image, blur_strength: int = 15, threshold: float = 0.5) -> Image.Image:
    """
    Applies segmentation using the RMBG-2.0 model and composites the original image with
    a Gaussian-blurred background based on an adjustable mask sensitivity threshold.
    """
    image = input_image.convert("RGB")
    orig_width, orig_height = image.size

    # Preprocess image for segmentation (resize only for model inference)
    input_tensor = segmentation_transform(image).unsqueeze(0).to(device)
    
    with torch.no_grad():
        preds = segmentation_model(input_tensor)[-1].sigmoid().cpu()
    pred = preds[0].squeeze()
    
    # Create binary mask with adjustable threshold (mask sensitivity)
    binary_mask = (pred > threshold).float()
    mask_pil = transforms.ToPILImage()(binary_mask).convert("L")
    mask_pil = mask_pil.point(lambda p: 255 if p > 128 else 0)
    mask_pil = mask_pil.resize((orig_width, orig_height), resample=Image.BILINEAR)
    
    blurred_image = image.filter(ImageFilter.GaussianBlur(blur_strength))
    final_image = Image.composite(image, blurred_image, mask_pil)
    return final_image

def depth_based_lens_blur(input_image: Image.Image, max_blur: float = 2, num_bands: int = 40, invert_depth: bool = False) -> Image.Image:
    """
    Applies a depth-based blur effect using a depth map produced by Depth-Anything.
    The effect simulates a lens blur where the max_blur parameter controls the maximum blur.
    This function uses the original input image size.
    """
    # Use the original image for depth estimation (no resizing)
    image_original = input_image.convert("RGB")
    
    # Obtain depth map using the pipeline (assumes model accepts variable sizes)
    results = depth_pipeline(image_original)
    depth_map_image = results['depth']
    
    depth_array = np.array(depth_map_image, dtype=np.float32)
    d_min, d_max = depth_array.min(), depth_array.max()
    depth_norm = (depth_array - d_min) / (d_max - d_min + 1e-8)
    if invert_depth:
        depth_norm = 1.0 - depth_norm

    orig_rgba = image_original.convert("RGBA")
    final_image = orig_rgba.copy()
    
    band_edges = np.linspace(0, 1, num_bands + 1)
    for i in range(num_bands):
        band_min = band_edges[i]
        band_max = band_edges[i + 1]
        mid = (band_min + band_max) / 2.0
        blur_radius_band = (1 - mid) * max_blur
        
        blurred_version = orig_rgba.filter(ImageFilter.GaussianBlur(blur_radius_band))
        band_mask = ((depth_norm >= band_min) & (depth_norm < band_max)).astype(np.uint8) * 255
        band_mask_pil = Image.fromarray(band_mask, mode="L")
        final_image = Image.composite(blurred_version, final_image, band_mask_pil)
    
    return final_image.convert("RGB")

def process_image(input_image: Image.Image, effect: str, mask_sensitivity: float, blur_strength: float) -> Image.Image:
    """
    Applies the selected effect:
      - "Gaussian Blur Background": uses segmentation with adjustable mask sensitivity and blur strength.
      - "Depth-based Lens Blur": applies depth-based blur with an adjustable maximum blur.
    """
    if effect == "Gaussian Blur Background":
        return segment_and_blur_background(input_image, blur_strength=int(blur_strength), threshold=mask_sensitivity)
    elif effect == "Depth-based Lens Blur":
        return depth_based_lens_blur(input_image, max_blur=blur_strength)
    else:
        return input_image

# ----------------------------
# Gradio Blocks Layout
# ----------------------------

with gr.Blocks(title="Interactive Blur Effects Demo") as demo:
    gr.Markdown(
        """
        # Interactive Blur Effects Demo
        Upload an image and choose an effect below.  
        For **Gaussian Blur Background**, adjust the mask sensitivity (controls segmentation threshold) 
        and blur strength (controls Gaussian blur radius).  
        For **Depth-based Lens Blur**, the blur strength slider sets the maximum blur intensity.
        """
    )
    
    with gr.Row():
        with gr.Column():
            input_image = gr.Image(type="pil", label="Input Image")
            effect_choice = gr.Radio(
                choices=["Gaussian Blur Background", "Depth-based Lens Blur"],
                label="Select Effect",
                value="Gaussian Blur Background"
            )
            mask_sensitivity_slider = gr.Slider(
                minimum=0.0, maximum=1.0, value=0.5, step=0.01,
                label="Mask Sensitivity (for segmentation)"
            )
            blur_strength_slider = gr.Slider(
                minimum=0, maximum=30, value=15, step=1,
                label="Blur Strength"
            )
            run_button = gr.Button("Apply Effect")
        with gr.Column():
            output_image = gr.Image(type="pil", label="Output Image")
    
    run_button.click(
        fn=process_image,
        inputs=[input_image, effect_choice, mask_sensitivity_slider, blur_strength_slider],
        outputs=output_image
    )

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