minor bugfixes (#125)

* minor bugfixes

* improved logging to debug missing classification images

* improved logging and no classification due to value below threshold

* Upgraded version

Author: stefanodellosa-personal

requirements.txt

@@ -17,5 +17,5 @@
 # Date: 2024-08-14
 # Description: module to keep track of WADAS version
 
-__version__ = "v0.5.1"
+__version__ = "v0.5.2"
 __dbversion__ = __version__

wadas/_version.py

@@ -109,7 +109,6 @@ def classify(self, img, results, classification_threshold):
             # Performing classification
             classification_result = self.classify_crop(cropped_image, classification_threshold)
             if classification_result[0]:
-
                 classified_animals.append(
                     {
                         "id": classification_id,
@@ -129,6 +128,7 @@ def classify_crop(self, crop_img, classification_threshold):
         labels = txt_animalclasses[self.language]
 
         if max(logits) < classification_threshold:
+            logger.info("Classification value under selected threshold.")
             return ["", 0]
 
         return [labels[np.argmax(logits)], max(logits)]

wadas/ai/pipeline.py

@@ -145,21 +145,28 @@ def classify(self, img_path, results):
         logger.info("Running classification on %s image...", img_path)
         img = Image.open(img_path).convert("RGB")
 
-        classified_animals = self.detection_pipeline.classify(
-            img, results, AiModel.classification_threshold
-        )
-
-        for detection in classified_animals:
-            # Cropping detection result(s) from original image leveraging detected boxes
-            cropped_image = img.crop(detection["xyxy"])
-            cropped_image_path = os.path.join(
-                "classification_output", f"{detection['id']}_cropped_image.jpg"
+        classified_animals = None
+        classified_img_path = None
+        if not (
+            classified_animals := self.detection_pipeline.classify(
+                img, results, AiModel.classification_threshold
             )
-            cropped_image.save(cropped_image_path)
-            logger.debug("Saved crop of image at %s.", cropped_image_path)
-
-        img_path = self.build_classification_square(img, classified_animals, img_path)
-        return img_path, classified_animals
+        ):
+            logger.debug("No classified animals, skipping img crops saving.")
+        else:
+            for classified_animal in classified_animals:
+                # Cropping detection result(s) from original image leveraging detected boxes
+                cropped_image = img.crop(classified_animal["xyxy"])
+                cropped_image_path = os.path.join(
+                    "classification_output", f"{classified_animal['id']}_cropped_image.jpg"
+                )
+                cropped_image.save(cropped_image_path)
+                logger.debug("Saved crop of image at %s.", cropped_image_path)
+
+            classified_img_path = self.build_classification_square(
+                img, classified_animals, img_path
+            )
+        return classified_img_path, classified_animals
 
     def build_classification_square(self, img, classified_animals, img_path):
         """Build square on classified animals."""

wadas/domain/ai_model.py

@@ -78,7 +78,7 @@ def run(self):
                                 f"animal from camera {cur_img['img_id']}!"
                             )
                         else:
-                            logger.info("No animals to classify.")
+                            logger.info("No animal classified.")
                             message = ""
                     else:
                         message = "WADAS has detected an animal from camera %s!" % id

wadas/domain/animal_detection_mode.py

@@ -96,6 +96,6 @@ def run(self):
                                 self.last_classified_animals_str,
                             )
                     else:
-                        logger.info("No animals to classify.")
+                        logger.info("No animal classified.")
 
         self.execution_completed()

wadas/domain/custom_classification_mode.py

@@ -160,6 +160,8 @@ def _classify(self, detection_event: DetectionEvent):
                 # Update detection event into db, if enabled
                 if db := DataBase.get_enabled_db():
                     db.update_detection_event(detection_event)
+            else:
+                logger.debug("No classified animals or classification results below threshold.")
 
     def ftp_camera_exist(self):
         """Method that returns True if at least an FTP camera exists, False otherwise."""

wadas/domain/operation_mode.py

@@ -83,7 +83,7 @@ def run(self):
                     f"animal from camera {detection_event.classification_img_path}!"
                 )
             else:
-                logger.info("No animals to classify.")
+                logger.info("No animal classified.")
                 message = ""
 
         # Send notification

wadas/domain/test_model_mode.py

@@ -34,8 +34,8 @@
     QWidget,
 )
 
-from ui.error_message_dialog import WADASErrorMessage
 from wadas.domain.database import DataBase, DBUser
+from wadas.ui.error_message_dialog import WADASErrorMessage
 from wadas.ui.qt.ui_configure_web_interface import Ui_DialogConfigureWebInterface
 
 module_dir_path = os.path.dirname(os.path.abspath(__file__))