app.py

from transformers import CLIPSegProcessor, CLIPSegForImageSegmentation
import gradio as gr
from PIL import Image
import torch
import numpy as np
import threading

processor = CLIPSegProcessor.from_pretrained("CIDAS/clipseg-rd64-refined")
model = CLIPSegForImageSegmentation.from_pretrained("CIDAS/clipseg-rd64-refined")

# Function to process image and generate mask
# Function to process image and generate mask
def process_image(image, prompt):
    inputs = processor(
        text=prompt, images=image, padding="max_length", return_tensors="pt"
    )

    # Extract image tensor and normalize it
    image_tensor = inputs["pixel_values"].squeeze().permute(1, 2, 0).cpu().numpy()
    image_tensor = (image_tensor * 255).astype(np.uint8)
    image_tensor = Image.fromarray(image_tensor)
    image_tensor = image_tensor.convert("RGB")
    
    # Perform CLIPSeg processing
    inputs = processor(
        text=prompt, images=image_tensor, padding="max_length", return_tensors="pt"
    )
    with torch.no_grad():
        outputs = model(**inputs)
        preds = outputs.logits

    pred = torch.sigmoid(preds)
    mat = pred.cpu().numpy()
    mask = Image.fromarray(np.uint8(mat * 255), "L")
    mask = mask.convert("RGB")
    mask = mask.resize(image.size)
    mask = np.array(mask)[:, :, 0]

    mask_min = mask.min()
    mask_max = mask.max()
    mask = (mask - mask_min) / (mask_max - mask_min)

    return mask


# Function to extract image using positive and negative prompts
def extract_image(pos_prompts, neg_prompts, img, threshold):
    positive_masks = get_masks(pos_prompts, img, 0.5)
    negative_masks = get_masks(neg_prompts, img, 0.5)

    pos_mask = np.any(np.stack(positive_masks), axis=0)
    neg_mask = np.any(np.stack(negative_masks), axis=0)
    final_mask = pos_mask & ~neg_mask

    final_mask = Image.fromarray(final_mask.astype(np.uint8) * 255, "L")
    output_image = Image.new("RGBA", img.size, (0, 0, 0, 0))
    output_image.paste(img, mask=final_mask)

    return output_image, final_mask

# Define Gradio interface
iface = gr.Interface(
    fn=extract_image,
    inputs=[
        gr.Textbox(
            label="Please describe what you want to identify (comma separated)",
            key="pos_prompts",
        ),
        gr.Textbox(
            label="Please describe what you want to ignore (comma separated)",
            key="neg_prompts",
        ),
        gr.Image(type="pil", label="Input Image", key="img"),
        gr.Slider(minimum=0, maximum=1, default=0.4, label="Threshold", key="threshold"),
    ],
    outputs=[
        gr.Image(label="Result", key="output_image"),
        gr.Image(label="Mask", key="output_mask"),
    ],
)

# Launch Gradio API
iface.launch()