Speaker
Description
"This study introduces a new method for monitoring oil spills and ships using Synthetic Aperture Radar (SAR) raw data and deep learning techniques. The proposed approach involves several key steps: pre-processing (including focusing, filtering, and land-sea mask), semantic segmentation, and classification using a deep convolutional neural network (DCNN) model. Real-time processing based on FFT ensures rapid response times.
For training the DCNN model, three datasets were combined: CleanSeaNet, TenGeoP-SARwv, and GAP_OilSpill_DB. The first two datasets are publicly available, while the authors created the third dataset by integrating documented case studies from news articles and cases identified in the sea area near the port of Brindisi, validated by expert GAP operators.
Data augmentation techniques were employed to enhance the model's performance by generating additional training data. The DCNN model utilizes DeepLab v3+ based on ResNet-18 architecture and is trained on a large SAR image dataset that includes various types of oil spills, look-alikes, novelty objects, and ships.
The proposed system is optimized for on-board satellite processing, ensuring real-time responses. Images are transmitted to the ground segment only when events of interest occur, such as the detection of novelty objects or potential oil spills involving nearby ships.
The study demonstrates that this approach offers a promising solution for real-time monitoring of oil spills, ships, and novelty objects using satellite SAR raw data. The integration of deep learning and data augmentation techniques significantly enhances detection accuracy and speed, leading to improved environmental management and oil spill response. Furthermore, this approach can be applied to diverse SAR datasets and has the potential for integration with existing oil spill response systems.
Acknowledgments
This work was carried out in the framework of the APP4AD project (“Advanced Payload data Processing for Autonomy & Decision”, Bando ASI “Tecnologie Abilitanti Trasversali”, Codice Unico di Progetto F95F21000020005), funded by the Italian Space Agency (ASI). ERS, ENVISAT and Sentinel-1 data are provided by the European Space Agency (ESA)."
