Update app.py

app.py

@@ -47,10 +47,10 @@ def segment_to_bbox(segment_indexs):
     return [np.min(x_points), np.min(y_points), np.max(x_points), np.max(y_points)]
 
 def clipseg_prediction(image):
-    inputs = processor(text=prompts, images=[image] * len(prompts), padding="max_length", return_tensors="pt")
+    inputs = clip_processor(text=prompts, images=[image] * len(prompts), padding="max_length", return_tensors="pt")
     # predict
     with torch.no_grad():
-        outputs = model(**inputs)
+        outputs = clip_model(**inputs)
     preds = outputs.logits.unsqueeze(1)
     # Setting threshold and classify the image contains vehicle or not
     flat_preds = torch.sigmoid(preds.squeeze()).reshape((preds.shape[0], -1))
@@ -71,9 +71,9 @@ def clipseg_prediction(image):
 
     # Vehicle checking
     if bbox_area(vehicle_bbox) > bbox_area(damage_bbox):
-        return True, bbox_normalization(damage_bbox)
+        return True, [bbox_normalization(damage_bbox)]
     else:
-        return False, []
+        return False, [[]]
      
 
 @torch.no_grad()
@@ -89,7 +89,7 @@ def foward_pass(image_input: np.ndarray, points: List[List[int]]) -> np.ndarray:
 
     outputs = model.forward(image_embeddings=cache_data[1], **inputs)
     masks = processor.image_processor.post_process_masks(
-        outputs.pred_masks.cpu(), inputs["original_sizes"].cpu(), inputs["reshaped_input_sizes"].cpu()
+        outputs.pred_masks.cpu(), inputs["original_sizes"].to(device), inputs["reshaped_input_sizes"].to(device)
     )
     masks = masks[0].squeeze(0).numpy().transpose(1, 2, 0)