app.py

import gradio as gr
from transformers import pipeline
from PIL import Image, ImageFilter
import numpy as np

pipe = pipeline(task="depth-estimation", model="LiheYoung/depth-anything-small-hf")

def blur_img(img, blur):
    depth = pipe(img)["depth"]
    depth_array = np.array(depth) / 255.0

    blurred_image_pil = img.filter(ImageFilter.BoxBlur(blur))
    blurred_image_array = np.array(blurred_image_pil)
    mask_array = np.expand_dims(depth_array, axis=-1)
    result_array = np.uint8(np.array(img) * mask_array + blurred_image_array * (1 - mask_array))
    result_pil = Image.fromarray(result_array)
    return result_pil
    
#interface
inputs = [
    gr.Image(type="pil", label="Input Image"),
    gr.Slider(minimum=1, maximum=10, label="Blur Amount", step=1)
]

output = gr.Image(type="pil", label="Blurred Image")

app= gr.Interface(fn=blur_img, inputs=inputs, outputs=output, title="Depth Anything model based Image Blurring", 
             description="Upload an image and adjust the blur amount using the slider. \nThis project takes in an image, obtains a depth mask using the Depth Anything model by https://huggingface.co/LiheYoung/depth-anything-base-hf, and blurs image based on the depth mask.", 
                 )

app.launch(share=True)