Update app.py

app.py

@@ -44,7 +44,7 @@ def get_masks(prompts, img, threhsold):
     return masks
 
 
-def extract_image(img, pos_prompts, neg_prompts, threshold):
+def extract_image(img, pos_prompts, neg_prompts, threshold, blur_radius=5):
     positive_masks = get_masks(pos_prompts, img, threshold)
     negative_masks = get_masks(neg_prompts, img, threshold)
 
@@ -53,6 +53,12 @@ def extract_image(img, pos_prompts, neg_prompts, threshold):
     neg_mask = np.any(np.stack(negative_masks), axis=0)
     final_mask = pos_mask & ~neg_mask
 
+    # apply Gaussian blur for feathering
+    final_mask_img = Image.fromarray((final_mask * 255).astype(np.uint8), "L")
+    final_mask_img = final_mask_img.filter(ImageFilter.GaussianBlur(radius=blur_radius))
+    final_mask = np.array(final_mask_img) / 255
+    final_mask = final_mask > threshold
+
     # extract the final image
     final_mask = Image.fromarray(final_mask.astype(np.uint8) * 255, "L")
     inverse_mask = np.invert(final_mask)
@@ -62,6 +68,7 @@ def extract_image(img, pos_prompts, neg_prompts, threshold):
     return output_image, final_mask, inverse_mask
 
 
+
 title = "Interactive demo: zero-shot image segmentation with CLIPSeg"
 description = "Demo for using CLIPSeg, a CLIP-based model for zero- and one-shot image segmentation. To use it, simply upload an image and add a text to mask (identify in the image), or use one of the examples below and click 'submit'. Results will show up in a few seconds."
 article = "<p style='text-align: center'><a href='https://arxiv.org/abs/2112.10003'>CLIPSeg: Image Segmentation Using Text and Image Prompts</a> | <a href='https://huggingface.co/docs/transformers/main/en/model_doc/clipseg'>HuggingFace docs</a></p>"