Abstract:Oil spills are threat to marine and coastal environments, causing extensive ecological damage and economic loss. Many advances have been made in the application of Machine Learning (ML) and Deep Learning (DL) techniques to detect oil spillage from satellite Synthetic Aperture Radar (SAR), or Unmanned Aerial Vehicle (UAV) images to overcome limitations of coverage, response time, and accuracy inherent in physical inspection method. This study leverages these technological advances by proposing a lightweight Convolutional Neural Network (CNN) architecture for fast detection and ensuring that the efficiency and precision of oil spill monitoring to facilitate the required response is not compromised. The CNN was trained on an image dataset of an offshore environment that captured both spillage and non-spillage conditions. The evaluation of the trained CNN model on test images results in classification accuracy of 93%.