Update app.py

app.py

@@ -8,6 +8,7 @@ from transformers import (
 from PIL import Image, ImageFilter
 import numpy as np
 import gradio as gr
+import cv2
 
 # Suppress specific warnings
 import warnings
@@ -51,7 +52,7 @@ def process_image(image):
     normalized_depth = (predicted_depth - min_depth) / (max_depth - min_depth)
     depth_map_image = Image.fromarray((normalized_depth * 255).astype(np.uint8))
     
-    # ------------------ Blurred Background Effect ------------------
+    # ------------------ Gaussian Blurred Background Effect ------------------
     # Invert the depth map
     inverted_depth = 1 - normalized_depth
     inverted_depth = (inverted_depth - inverted_depth.min()) / (inverted_depth.max() - inverted_depth.min())
@@ -62,17 +63,39 @@ def process_image(image):
     depth_weight_resized = np.expand_dims(depth_weight_resized, axis=-1)
     
     # Apply Gaussian blur to the entire image
-    blurred_image = image.filter(ImageFilter.GaussianBlur(radius=15))
+    gaussian_blurred_image = image.filter(ImageFilter.GaussianBlur(radius=15))
+    gaussian_blurred_np = np.array(gaussian_blurred_image).astype(np.float32)
     
-    # Convert images to numpy arrays
+    # Blend images based on the depth weight
     original_np = np.array(image).astype(np.float32)
-    blurred_np = np.array(blurred_image).astype(np.float32)
+    composite_gaussian_np = (1 - depth_weight_resized) * original_np + depth_weight_resized * gaussian_blurred_np
+    composite_gaussian_image = Image.fromarray(np.clip(composite_gaussian_np, 0, 255).astype(np.uint8))
+    
+    # ------------------ Lens Blurred Background Effect ------------------
+    # Convert PIL image to OpenCV format
+    original_cv = cv2.cvtColor(np.array(image), cv2.COLOR_RGB2BGR)
+    
+    # Apply Lens Blur using OpenCV's blur with a larger kernel
+    # Note: OpenCV does not have a direct lens blur function, but we can approximate it
+    # by using a larger kernel size. For a more realistic lens blur, additional processing is required.
+    lens_blur_kernel_size = 21  # Adjust kernel size for stronger blur
+    lens_blurred_cv = cv2.GaussianBlur(original_cv, (lens_blur_kernel_size, lens_blur_kernel_size), 0)
+    
+    # Convert back to PIL Image
+    lens_blurred_image = Image.fromarray(cv2.cvtColor(lens_blurred_cv, cv2.COLOR_BGR2RGB))
+    lens_blurred_np = np.array(lens_blurred_image).astype(np.float32)
     
     # Blend images based on the depth weight
-    composite_np = (1 - depth_weight_resized) * original_np + depth_weight_resized * blurred_np
-    composite_image = Image.fromarray(np.clip(composite_np, 0, 255).astype(np.uint8))
+    composite_lens_np = (1 - depth_weight_resized) * original_np + depth_weight_resized * lens_blurred_np
+    composite_lens_image = Image.fromarray(np.clip(composite_lens_np, 0, 255).astype(np.uint8))
+    
+    # Return results
+    binary_mask_image = binary_mask_image.convert("L")  # Ensure it's in grayscale
+    depth_map_image = depth_map_image.convert("L")      # Ensure it's in grayscale
+    gaussian_blurred_image = composite_gaussian_image
+    lens_blurred_image = composite_lens_image
     
-    return image, binary_mask_image, depth_map_image, composite_image
+    return image, binary_mask_image, depth_map_image, gaussian_blurred_image, lens_blurred_image
 
 # Define Gradio interface using the updated API
 interface = gr.Interface(
@@ -80,12 +103,13 @@ interface = gr.Interface(
     inputs=gr.Image(type="pil", label="Upload Image"),
     outputs=[
         gr.Image(type="pil", label="Original Image"),
-        gr.Image(type="pil", label="Segmentation Mask"),
+        gr.Image(type="pil", label="Segmentation Mask (B/W)"),
         gr.Image(type="pil", label="Depth Map"),
-        gr.Image(type="pil", label="Blurred Background Effect"),
+        gr.Image(type="pil", label="Gaussian Blurred Background"),
+        gr.Image(type="pil", label="Lens Blurred Background"),
     ],
-    title="Semantic Segmentation and Depth Estimation",
-    description="Upload an image to generate a segmentation mask, depth map, and a blurred background effect.",
+    title="Semantic Segmentation and Dual Blur Effects",
+    description="Upload an image to generate a segmentation mask, depth map, Gaussian blurred background, and lens blurred background effect.",
     examples=[
         ["Selfie_1.jpg"],
         ["Selfie_2.jpg"]