Related Experiment Video
Updated: Mar 28, 2026
Combining Eye-tracking Data with an Analysis of Video Content from Free-viewing a Video of a Walk in an Urban Park Environment
Published on: May 7, 2019
Offshore oil spill detection based on visual information and deep learning
Rasim Alguliyev1, Ramiz Aliguliyev1, Lyudmila Sukhostat1
1Institute of Information Technology, 9A, B. Vahabzade Street, Baku, AZ1141, Azerbaijan.
Abstract:
Timely detection of offshore oil spills that cause irreparable harm to aquatic life can reduce the risk of damage and ensure environmental safety. For this purpose, images obtained by synthetic aperture radar installed on satellites and aircraft are widely used to obtain a visual representation of the sea surface. In this paper, an approach consisting of two U-Net-based models using EfficientNet-B7 and ResNeXt that analyse images to detect oil spills is proposed. The evaluation of the proposed approach is conducted on PALSAR and Sentinel-1 datasets. The proposed model is compared with different U-Net backbone models and shows high results according to mean IoU and Dice scores, which are equal to 81.62% and 83.27% for the PALSAR dataset and 82.41% and 84.36% for the Sentinel-1 dataset. The research results demonstrate the high efficiency of oil spill detection.
