nodelib 0.17.1

detector/docker.sh

@@ -29,7 +29,7 @@ fi
 # ========================== BUILD CONFIGURATION / IMAGE SELECTION =======================
 
 SEMANTIC_VERSION=0.1.13
-NODE_LIB_VERSION=0.17.0
+NODE_LIB_VERSION=0.17.1
 build_args=" --build-arg NODE_LIB_VERSION=$NODE_LIB_VERSION"
 
 if [ -f /etc/nv_tegra_release ] # Check if we are on a Jetson device

detector/yolov5_detector.py

@@ -32,10 +32,12 @@ def __init__(self) -> None:
         self.conf_threshold = float(os.getenv('CONF_THRESHOLD', '0.2'))
 
     def init(self) -> None:
-        assert self.model_info is not None, 'model_info must be set before calling init()'
-        resolution = self.model_info.resolution
-        assert resolution is not None
-        engine_file = self._create_engine(resolution,
+        if self.model_info is None:
+            raise RuntimeError('Model info not initialized. Call load_model_info_and_init_model() first.')
+        if not isinstance(self.model_info.resolution, int) or self.model_info.resolution <= 0:
+            raise RuntimeError("resolution must be an integer > 0")
+
+        engine_file = self._create_engine(self.model_info.resolution,
                                           len(self.model_info.categories),
                                           self.model_info.model_size,
                                           f'{self.model_info.model_root_path}/model.wts')
@@ -87,8 +89,10 @@ def clip_point(x: float, y: float, img_width: int, img_height: int) -> Tuple[flo
         return x, y
 
     def evaluate(self, image: bytes) -> ImageMetadata:
-        assert self.yolov5 is not None, 'init() must be executed first. Maybe loading the engine failed?!'
-        assert self.model_info is not None, 'model_info must be set before calling evaluate()'
+        if self.yolov5 is None:
+            raise RuntimeError('init() must be executed first. Maybe loading the engine failed?!')
+        if self.model_info is None:
+            raise RuntimeError('model_info must be set before calling evaluate()')
 
         image_metadata = ImageMetadata()
 
@@ -130,9 +134,10 @@ def evaluate(self, image: bytes) -> ImageMetadata:
             if skipped_detections:
                 log_msg = '\n'.join([str(d) for d in skipped_detections])
                 self.log.warning('Removed %d small detections from result: \n%s', len(skipped_detections), log_msg)
-        except Exception:
-            self.log.exception('inference failed')
-        return image_metadata
+            return image_metadata
+
+        except Exception as e:
+            raise RuntimeError('Error during inference') from e
 
     def batch_evaluate(self, images: List[bytes]) -> ImagesMetadata:
         raise NotImplementedError('batch_evaluate is not implemented for Yolov5Detector')

detector_cpu/docker.sh

@@ -28,8 +28,8 @@ fi
 
 # ========================== BUILD CONFIGURATION / IMAGE SELECTION =======================
 
-SEMANTIC_VERSION=0.1.11
-NODE_LIB_VERSION=0.17.0
+SEMANTIC_VERSION=0.1.12
+NODE_LIB_VERSION=0.17.1
 build_args=" --build-arg NODE_LIB_VERSION=$NODE_LIB_VERSION"
 
 

detector_cpu/yolov5_detector.py

@@ -28,16 +28,20 @@ def __init__(self) -> None:
         self.conf_threshold = float(os.getenv('CONF_THRESHOLD', '0.2'))
 
     def init(self) -> None:
-        assert self.model_info is not None, 'model_info must be set before calling init()'
-        resolution = self.model_info.resolution
-        assert resolution is not None
-        pt_file = f'{self.model_info.model_root_path}/model.pt'
+        if self.model_info is None:
+            raise RuntimeError('Model info not initialized. Call load_model_info_and_init_model() first.')
+        if not isinstance(self.model_info.resolution, int) or self.model_info.resolution <= 0:
+            raise RuntimeError("model_info.resolution must be an integer > 0")
 
+        pt_file = f'{self.model_info.model_root_path}/model.pt'
         yolov5_path = os.path.join(
             os.path.dirname(__file__), 'app_code', 'yolov5')
         self.yolov5 = torch.hub.load(
             yolov5_path, 'custom', pt_file, source='local')
-        assert self.yolov5 is not None
+
+        if self.yolov5 is None:
+            raise RuntimeError('Failed to load YOLOv5 model')
+
         self.yolov5.eval()
 
     @staticmethod
@@ -76,11 +80,14 @@ def clip_point(x: float, y: float, img_width: int, img_height: int) -> Tuple[flo
         return x, y
 
     def evaluate(self, image: bytes) -> ImageMetadata:
-        assert self.yolov5 is not None, 'init() must be executed first!'
-        assert self.model_info is not None, 'model_info must be set before calling evaluate()'
+        if self.yolov5 is None:
+            raise RuntimeError('Model not initialized. Call load_model_info_and_init_model() first.')
+        if self.model_info is None:
+            raise RuntimeError('Model info not initialized. Call load_model_info_and_init_model() first.')
 
         image_metadata = ImageMetadata()
-        assert self.model_info.resolution, "input_size must be an integer > 0"
+        if not isinstance(self.model_info.resolution, int) or self.model_info.resolution <= 0:
+            raise RuntimeError("input_size must be an integer > 0")
         self.input_size = self.model_info.resolution
 
         try:
@@ -144,9 +151,8 @@ def evaluate(self, image: bytes) -> ImageMetadata:
             self.log.debug('took %f s', time.time() - t)
             return image_metadata
 
-        except Exception:
-            self.log.exception('inference failed')
-            return image_metadata
+        except Exception as e:
+            raise RuntimeError('Error during inference') from e
 
     def _preprocess_image(self, raw_bgr_image):
         """