fix checks

predicators/spot_utils/perception_utils.py

@@ -234,10 +234,8 @@ def query_detic_sam(image_in: np.ndarray, classes: List[str],
 #     return np.vstack((x, y, z)).T, valid_inds
 
 
-def process_image_response(
-        image_response: bosdyn.api.image_pb2.ImageResponse,
-        to_rgb: bool = False
-) -> np.ndarray:
+def process_image_response(image_response: bosdyn.api.image_pb2.ImageResponse,
+                           to_rgb: bool = False) -> np.ndarray:
     """Given a Boston Dynamics SDK image response, extract the correct np array
     corresponding to the image."""
     # pylint: disable=no-member
@@ -272,7 +270,6 @@ def process_image_response(
     else:
         img = cv2.imdecode(img, -1)
 
-    # TODO also add here; to check duplicate
     # Convert to RGB color, as some perception models assume RGB format
     # By default, still use BGR to keep backward compability
     if to_rgb:

predicators/spot_utils/spot_utils.py

@@ -284,12 +284,9 @@ def get_single_camera_image(self,
         img_req = build_image_request(
             source_name,
             quality_percent=100,
-            pixel_format=(
-                None if
-                ('hand' in source_name or 'depth' in source_name) else
-                image_pb2.Image.PIXEL_FORMAT_RGB_U8
-            )
-        )
+            pixel_format=(None if
+                          ('hand' in source_name or 'depth' in source_name)
+                          else image_pb2.Image.PIXEL_FORMAT_RGB_U8))
         image_response = self.image_client.get_image([img_req])
 
         # Format image before detecting apriltags.
@@ -493,15 +490,11 @@ def get_sam_object_loc_from_camera(
         """
         _, rgb_img_response = self.get_single_camera_image(source_rgb, True)
         _, depth_img_response = self.get_single_camera_image(
-            source_depth, True)
-        # TODO fix - to rgb in processing response
+            source_depth, False)
         image = {
-            'rgb': process_image_response(
-                rgb_img_response[0], to_rgb=True
-            ),
-            'depth': process_image_response(
-                depth_img_response[0], to_rgb=True
-            ),
+            'rgb': process_image_response(rgb_img_response[0], to_rgb=True),
+            'depth': process_image_response(depth_img_response[0],
+                                            to_rgb=False),
         }
         image_responses = {
             'rgb': rgb_img_response[0],
@@ -970,8 +963,7 @@ def arm_object_grasp(self, obj: Object) -> None:
             results = get_pixel_locations_with_detic_sam(
                 classes=[obj_name_to_vision_prompt['brush']],
                 in_res_image=image_for_sam,
-                plot=CFG.spot_visualize_vision_model_outputs
-            )
+                plot=CFG.spot_visualize_vision_model_outputs)
 
             if len(results) > 0:
                 # We only want the most likely sample (for now).