fix to_tuple error and force sam on cuda

blurnonymize.py

@@ -152,13 +152,12 @@ class ImageBlurnonymizer:
         self.init_sam()
 
     def init_sam(self, force=False):
-        return
         # only initialize SAM if it hasn't been initialized yet
         if self.predictor is not None and not force:
             return
 
         # self.device = "cuda" if torch.cuda.is_available() else "cpu"
-        self.device = "cpu"
+        self.device = "cuda"
         self.predictor = SAM2ImagePredictor.from_pretrained("facebook/sam2.1-hiera-small", device=self.device)
 
     @staticmethod
@@ -244,8 +243,7 @@ class ImageBlurnonymizer:
         # Ensure points are valid coordinates (e.g., non-negative)
         points = [[max(0, p[0]), max(0, p[1])] for p in points]
 
-        # with torch.inference_mode(), torch.autocast(self.device, dtype=torch.bfloat16):
-        with torch.inference_mode(), torch.autocast(self.device):
+        with torch.inference_mode(), torch.autocast(self.device, dtype=torch.bfloat16):
             self.predictor.set_image(image)
             masks, scores, _ = self.predictor.predict(
                 box=np.array(bbox),  # Predictor might expect numpy array

utils.py

@@ -88,6 +88,10 @@ class BoundingBox(BaseModel):
         """Creates a BoundingBox instance from a label and a list of coordinates."""
         return BoundingBox(label=label, x_min=box[0], y_min=box[1], x_max=box[2], y_max=box[3])
 
+    def to_tuple(self) -> tuple[int, int, int, int]:
+        """Converts the BoundingBox instance to a tuple of coordinates."""
+        return (self.x_min, self.y_min, self.x_max, self.y_max)
+
 def parse_json_response(out: str) -> list[dict]:
     """Extracts and parses JSON content from a string.