This research employs machine learning to analyze Sentinel-1 satellite imagery for ship detection in Saudi Arabian petroleum ports and their anchorage zones, aiming to correlate maritime activity with the Organization of the Petroleum Exporting Countries (OPEC) monthly reports on oil supply and production figures specific to Saudi Arabia. The goal is to understand global oil supply dynamics through maritime surveillance analytics.
The repository includes scripts for data acquisition, pre-processing of the OpenSARShip dataset and collecting Sentinel-1 images. Spectral pre-processing tools will cover image enhancement, noise reduction, and spectral calibration. The repository will also contain the ship detection model, which leverages a convolutional neural network (CNN) with optimized architecture for object detection, along with hyperparameter tuning scripts.