Oil spills pose a serious threat to marine ecosystems, affecting aquatic biodiversity, coastal environments, and human livelihoods. Traditional detection methods relying on satellite images, radar, or manual observation often suffer from delays and limited accuracy. This project presents a machine learning-based approach for oil spill detection using remote sensing imagery, particularly satellite, Synthetic Aperture Radar (SAR), and hyperspectral data. The system uses Convolutional Neural Networks (CNNs) to analyze and extract features from images, accurately distinguishing oil spills from lookalike substances such as algae or natural films. Trained on labeled datasets of past incidents, the model identifies spills and predicts their possible spread, enabling faster intervention. By automating image-based analysis, this approach offers a scalable, efficient, and cost-effective solution for real-time marine pollution monitoring. Future enhancements may include drone surveillance integration and adaptive response using reinforcement learning, further supporting sustainable ocean protection.