Deep Learning Models for Oil Spill Detection in Marine Settings: A Literature Review

Abstract

Oil spills in marine settings can be identified and tracked by remote sensing. The accuracy and effectiveness of oil spill detection using faraway sensing data have shown tremendous promise for deep learning (DL) algorithms, particularly deep neural networks (DNNs). In this literature review, we summarized the key DL models that have been used in oil spill detection, including CNN, RNN, DBN, AE, and GAN. We also discussed the different components and tasks involved in DL models, such as pooling layers, forward and backpropagation, and optimization of weights. Additionally, we present several case studies that have successfully applied DL approach in oil spill recognition, including the use of DBN to differentiate oil spills from lookalikes in SAR images, and the use of spatial-spectral jointed SAE to acquire and categorize oil slicks on the ocean surface using hyperspectral data. The findings from these studies demonstrate the potential of DL models to improve the accuracy and proficiency of oil spill detection using RS data.