NOVEL ANNOTATION AND METRICS FOR MANGROVE SPECIES CLASSIFICATION USING BOUNDING BOX OBJECT DETECTION

Bounding boxes are predominant annotation and detection methods in object detection, and their accuracies are commonly gauged using the intersection-over-union (IoU) index and its derivatives. While these methods are effective for clearly defined objects, modifications are required when using bounding box classifiers with ambiguously defined objects. This study proposed novel annotation and metrics methods using bounding boxes for YOLOv5 to classify mangrove species in Unmanned Aerial Vehicle (UAV) imagery. Annotation was performed on each estimated canopy, while accuracy metrics were computed based on pixel intersection between detections and masks generated from the merged annotation on individual images in the validation dataset. Overall, the annotation method did not affect the training process, as the model produced promising results, and the computed metrics were useful in gauging model accuracy based on specific requirements. Future work may test the proposed annotation and metrics method on other similarly novel applications.