eval task save avg_time_per_images as result

openvinotoolkit · Oct 19, 2023 · df41ae6 · df41ae6
1 parent ab6a63e
commit df41ae6
Show file tree

Hide file tree

Showing 2 changed files with 33 additions and 0 deletions.
diff --git a/otx/algorithms/classification/adapters/openvino/task.py b/otx/algorithms/classification/adapters/openvino/task.py
@@ -20,6 +20,7 @@
 import os
 import tempfile
 import warnings
+import time
 from typing import Any, Dict, Optional, Tuple, Union
 from zipfile import ZipFile
 
@@ -175,6 +176,11 @@ def __init__(self, task_environment: TaskEnvironment):
         self.hparams = self.task_environment.get_hyper_parameters(ClassificationConfig)
         self.model = self.task_environment.model
         self.inferencer = self.load_inferencer()
+        self._avg_time_per_image = None
+
+    @property
+    def avg_time_per_image(self):
+        return self._avg_time_per_image
 
     def load_inferencer(self) -> ClassificationOpenVINOInferencer:
         """load_inferencer function of ClassificationOpenVINOTask."""
@@ -206,7 +212,9 @@ def infer(
             explain_predicted_classes = inference_parameters.explain_predicted_classes
 
         dataset_size = len(dataset)
+        total_time = 0.0
         for i, dataset_item in enumerate(dataset, 1):
+            start_time = time.perf_counter()
             predicted_scene, probs, saliency_map, repr_vector, act_score = self.inferencer.predict(dataset_item.numpy)
             item_labels = predicted_scene.annotations[0].get_labels()
             dataset_item.append_labels(item_labels)
@@ -235,7 +243,16 @@ def infer(
                         "Could not find Feature Vector and Saliency Map in OpenVINO output. "
                         "Please rerun OpenVINO export or retrain the model."
                     )
+            end_time = time.perf_counter() - start_time
+            total_time += end_time
             update_progress_callback(int(i / dataset_size * 100))
+
+
+        self._avg_time_per_image = total_time / dataset_size
+        logger.info(f"Avg time per image: {self._avg_time_per_image} secs")
+        logger.info(f"Total time: {total_time} secs")
+        logger.info("Classification OpenVINO inference completed")
+
         return dataset
 
     def explain(

diff --git a/otx/algorithms/segmentation/adapters/openvino/task.py b/otx/algorithms/segmentation/adapters/openvino/task.py
@@ -18,6 +18,7 @@
 import json
 import os
 import tempfile
+import time
 from typing import Any, Dict, Optional, Tuple, Union
 from zipfile import ZipFile
 
@@ -168,6 +169,7 @@ def __init__(self, task_environment: TaskEnvironment):
         self.model = self.task_environment.model
         self.model_name = self.task_environment.model_template.model_template_id
         self.inferencer = self.load_inferencer()
+        self._avg_time_per_image = None
 
         labels = task_environment.get_labels(include_empty=False)
         self._label_dictionary = dict(enumerate(labels, 1))
@@ -179,6 +181,10 @@ def hparams(self):
         """Hparams of OpenVINO Segmentation Task."""
         return self.task_environment.get_hyper_parameters(SegmentationConfig)
 
+    @property
+    def avg_time_per_image(self):
+        return self._avg_time_per_image
+
     def load_inferencer(self) -> OpenVINOSegmentationInferencer:
         """load_inferencer function of OpenVINO Segmentation Task."""
         if self.model is None:
@@ -204,7 +210,9 @@ def infer(
             process_soft_prediction = False
 
         dataset_size = len(dataset)
+        total_time = 0.0
         for i, dataset_item in enumerate(dataset, 1):
+            start_time = time.perf_counter()
             predicted_scene, feature_vector, soft_prediction = self.inferencer.predict(dataset_item.numpy)
             dataset_item.append_annotations(predicted_scene.annotations)
 
@@ -231,8 +239,16 @@ def infer(
                     )
                     dataset_item.append_metadata_item(result_media, model=self.model)
 
+            end_time = time.perf_counter() - start_time
+            total_time += end_time
             update_progress_callback(int(i / dataset_size * 100), None)
 
+
+        self._avg_time_per_image = total_time / dataset_size
+        logger.info(f"Avg time per image: {self._avg_time_per_image} secs")
+        logger.info(f"Total time: {total_time} secs")
+        logger.info("Segmentation OpenVINO inference completed")
+
         return dataset
 
     def evaluate(self, output_resultset: ResultSetEntity, evaluation_metric: Optional[str] = None):