Update app.py

app.py

@@ -3,46 +3,45 @@ import gradio as gr
 from PIL import Image
 import torch
 import matplotlib.pyplot as plt
-import cv2
+import numpy as np
 
 processor = CLIPSegProcessor.from_pretrained("CIDAS/clipseg-rd64-refined")
 model = CLIPSegForImageSegmentation.from_pretrained("CIDAS/clipseg-rd64-refined")
 
 def process_image(image, prompt):
-  inputs = processor(text=prompt, images=image, padding="max_length", return_tensors="pt")
-  
-  # predict
-  with torch.no_grad():
-    outputs = model(**inputs)
-    preds = outputs.logits
-  
-  filename = f"mask.png"
-  plt.imsave(filename, torch.sigmoid(preds))
-  
-  # # img2 = cv2.imread(filename)
-  # # gray_image = cv2.cvtColor(img2, cv2.COLOR_BGR2GRAY)
+    # Prepare inputs with the processor
+    inputs = processor(text=prompt, images=image, return_tensors="pt")
+    
+    # Predict
+    with torch.no_grad():
+        outputs = model(**inputs)
+        preds = outputs.logits.squeeze()  # Assuming the output logits is of shape [1, H, W]
+    
+    # Apply sigmoid to convert logits to probabilities
+    preds = torch.sigmoid(preds)
+    
+    # Convert to numpy array
+    mask = preds.numpy()
+    
+    # Save the image correctly handling dimensions
+    filename = "mask.png"
+    plt.imsave(filename, mask, cmap='gray')  # Use cmap='gray' for grayscale image saving
+    
+    # Convert to PIL Image and return
+    return Image.open(filename).convert("RGB")
 
-  # # (thresh, bw_image) = cv2.threshold(gray_image, 100, 255, cv2.THRESH_BINARY)
-  
-  # # # fix color format
-  # # cv2.cvtColor(bw_image, cv2.COLOR_BGR2RGB)
-  
-  # # return Image.fromarray(bw_image)
-
-  return Image.open("mask.png").convert("RGB")
-  
 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>"
+description = "Demo for using CLIPSeg, a CLIP-based model for zero- and one-shot image segmentation."
+article = "<p style='text-align: center'><a href='https://arxiv.org/abs/2112.10003'>CLIPSeg: Image Segmentation Using Text and Image Prompts</a></p>"
+
+examples = [["example_image.png", "a description of what to segment"]]
 
-examples = [["example_image.png", "wood"]]
-   
 interface = gr.Interface(fn=process_image, 
-                     inputs=[gr.Image(type="pil"), gr.Textbox(label="Please describe what you want to identify")],
-                     outputs=gr.Image(type="pil"),
-                     title=title,
-                     description=description,
-                     article=article,
-                     examples=examples)
-                     
-interface.launch(debug=True)
+                         inputs=[gr.Image(type="pil"), gr.Textbox(label="Please describe what you want to identify")],
+                         outputs=gr.Image(type="pil"),
+                         title=title,
+                         description=description,
+                         article=article,
+                         examples=examples)
+                         
+interface.launch(debug=True)