Hydrocarbon contamination in aquatic environments poses a major environmental challenge, which requires the development of advanced tools for its early detection. In this chapter, hyperspectral images captured with high spatial resolution local cameras are presented to identify the presence of hydrocarbons in water. Two main approaches are compared: classifiers trained with hyperspectral signatures using machine learning techniques and classifiers trained with full hyperspectral images (HSI) using deep learning methods, especially convolutional neural networks (CNNs). The comparative analysis provides critical insights into the performance of both methodologies, outlining their strengths and limitations in terms of accuracy, robustness, and generalization. The findings underscore the potential of deep learning approaches for improving the automated detection of hydrocarbon contamination, while also recognising the reliability of classical machine learning techniques in specific scenarios. This study advances the development of more efficient monitoring strategies for aquatic ecosystems, reinforcing the role of artificial intelligence in hyperspectral data analysis for environmental protection.