handle boxes and points for SAM input

handler.py

@@ -62,8 +62,15 @@ class EndpointHandler():
         # Process the image
         img = Image.open(io.BytesIO(image_bytes)).convert("RGB")
         
-        # Prepare inputs for the model
-        inputs = self.processor(img, return_tensors="pt")
+        # SAM requires input prompts, so we'll generate a center point prompt
+        height, width = img.size[1], img.size[0]  # PIL returns (width, height)
+        
+        # Create a center point prompt for automatic segmentation
+        input_points = [[[width // 2, height // 2]]]  # Center point
+        input_labels = [[1]]  # Positive prompt
+        
+        # Prepare inputs for the model with prompts
+        inputs = self.processor(img, input_points=input_points, input_labels=input_labels, return_tensors="pt")
         
         # Generate masks using the model
         with torch.no_grad():
@@ -76,9 +83,17 @@ class EndpointHandler():
             inputs["reshaped_input_sizes"].cpu()
         )[0]
         
-        # Convert the first mask to a binary mask
-        mask = masks[0].squeeze().numpy()
-        mask_binary = (mask > 0.0).astype(np.uint8) * 255
+        # Convert the best mask to a binary mask
+        # SAM returns multiple masks, take the first one
+        if len(masks) > 0:
+            mask = masks[0].squeeze().numpy()
+            mask_binary = (mask > 0.5).astype(np.uint8) * 255
+        else:
+            # Fallback: create a simple center mask
+            mask_binary = np.zeros((height, width), dtype=np.uint8)
+            center_x, center_y = width // 2, height // 2
+            size = min(width, height) // 8
+            mask_binary[center_y-size:center_y+size, center_x-size:center_x+size] = 255
         
         # Convert result to base64
         out = io.BytesIO()