Can Deep Learning Help Clean up Our Environment?
25 May 2021
The importance of protecting life under sea is reflected in the UN’s sustainable development goal (SDG) 14 which focuses on our need to conserve and sustainably use the oceans, seas and marine resources for sustainable development. Oil spills have significant detrimental impact on the environment, and specifically for the marine environment cause pollution to sea surfaces and harm bird species, fish and other aquatic life. Consequently, any technologies which can identify, and aid the removal of oil slicks, can make an important contribution to maintaining a safe and clean environment and protecting aquatic life to contribute to achieving SDG 14. A team of researchers, which included Professors Reem Salim, Hadil Abu Khalifeh, Adel Khelifi and Mohammed Ghazal from the College of Engineering with colleagues from the center of excellence BIO-Imaging Research Lab at the University of Louisville, use synthetic aperture radar sensors, which can be mounted on aircraft or satellites, to obtain images of sea and land surfaces which allows for identification of oil spills.
Their two-stage approach first uses convolutional Neural Network followed by five-stage U-Net to identify oil-spills. The proposed framework provides improved precision to identify oil-spills and to exclude other look-alike phenomena. Consequently, compared to current state-of-the-art approaches this improvement in precision allows valuable resources to be directed toward cleaning up pollution, while avoiding unnecessary look-alike oil-spills. Professor Mohamed Ghazal, Professor and Chair of the Electrical and Computer Engineering Department at ADU said, “This research project provides important insights which can help the environment, and are particularly important to the UAE as a global leader in the production and transportation of oil
Shaban, M., Salim, R., Khalifeh, H.A., Khelifi, A., Shalaby, A., El-Mashad, S., Mahmoud, A., Ghazal, M., El-Baz, A. (2021) A Deep-Learning Framework for the Detection of Oil Spills from SAR Data, Sensors 21(7). https://bit.ly/3yHt8BS
Impact Factor: 2.497; Scopus percentile: 87th.