A company that operates oil platforms uses drones to photograph locations on oil platforms that are difficult for humans to access to search for corrosion. Experienced engineers review the photos to d

An object detection algorithm is essential for identifying multiple, distinct corrosion areas within a single photo, providing bounding box locations for each instance of corrosion.

Amazon Rekognition with label detection, particularly Custom Labels, can be trained to identify specific types or characteristics of corrosion as labels within the detected areas or across the image, providing valuable features for subsequent severity classification.

K-means clustering is an unsupervised algorithm used for grouping similar data points, which is not suitable for a supervised classification task where the goal is to assign predefined severity categories.

An XGBoost algorithm is a robust supervised machine learning model well-suited for classifying the severity of corrosion into predefined categories (immediate, future, no action) based on features extracted from the detected corrosion areas and labels.

While performing image augmentation on photos that contain corrosion is a good practice for imbalanced datasets, it is a data preprocessing technique rather than a core model building or classification step, and other steps provide more direct functionality for detection and classification.

Performing image augmentation on photos that do not contain corrosion would further exacerbate the existing class imbalance, making it harder for the model to learn to identify the rare corrosion instances.