SAR Oil Spill Detector

SAR Oil Spill Detector

The SAR Oil Spill Detector is a tool deployed by M4D group to identify oil pollution incidents on the sea surface. The developed tool processes SAR image representations and produces annotated masks where detected oil slicks are highlighted accordingly.

Τhe detector is the key outcome of our recent research work [1] focused on employing deep learning architectures for the task of oil spill detection. Considering this task as an image semantic segmentation process to provide the capacity of estimating the size of dispersion, our tool utilizes DeepLabv3+ [2], a deep-learning model designed for this purpose. The model was trained on a developed dataset containing SAR images depicting oil spills and can segment instances of totally 5 classes, namely sea surface (black), oil spill (cyan), look-alike (red), ship (brown) and land (green). Apart from the offline testing, the deployed tool has also been validated through near real-life experiments to the extent possible.

The Oil Spill Detector can be deployed and tested locally by the user in a standalone manner. For further information please feel free to contact us.

System requirements
Supported OS: Ubuntu/Windows
Processor: Any CPU with 4 cores or more
RAM: 8 GigaBytes (GB) or more
Graphics card: NVIDIA GPU with 8GB memory or more

Contact information
Marios Krestenitis (mikretenitis@iti.gr)
Konstantinos Ioannidis (kioannid@iti.gr)
Stefanos Vrochidis (stefanos@iti.gr)

Data
“Oil Spill Detection Dataset” deployed by M4D and available to the community.

References
1. Krestenitis, M., Orfanidis, G., Ioannidis, K., Avgerinakis, K., Vrochidis, S., & Kompatsiaris, I. (2019). Oil spill identification from satellite images using deep neural networks. Remote Sensing, 11(15), 1762.
2. Chen, L. C., Zhu, Y., Papandreou, G., Schroff, F., & Adam, H. (2018). Encoder-decoder with atrous separable convolution for semantic image segmentation. In Proceedings of the European conference on computer vision (ECCV) (pp. 801-818).