Update app.py

app.py

@@ -9,26 +9,25 @@ model = CLIPModel.from_pretrained("openai/clip-vit-base-patch32")
 processor = CLIPProcessor.from_pretrained("openai/clip-vit-base-patch32")
 
 # Define a list of target words for the game
-words = ["cat", "car", "tree", "house", "dog"]  
-
+words = ["cat", "car", "tree", "house", "dog"]  # Add more words as needed
 
+# Precompute text embeddings for faster comparisons
 text_inputs = processor(text=words, return_tensors="pt", padding=True)
 with torch.no_grad():
     text_features = model.get_text_features(**text_inputs)
 
-
+# Define the function to process drawing and make a prediction
 def guess_drawing(drawing):
-   
-    drawing_data = drawing['data']  
-    image_array = np.array(drawing_data, dtype=np.uint8)  
-    
-    
+    # Assuming `drawing` is provided as an RGB or grayscale array
+    image_array = np.array(drawing, dtype=np.uint8)  # Directly convert it to a NumPy array
+
+    # Convert to PIL image
     image = Image.fromarray(image_array)
     
-    
+    # Prepare the image for the model
     image_inputs = processor(images=image, return_tensors="pt")
     
-    
+    # Get image features from the model
     with torch.no_grad():
         image_features = model.get_image_features(**image_inputs)
 
@@ -49,4 +48,3 @@ interface = gr.Interface(
 )
 
 interface.launch()
-