Mars is the most Earth-like terrestrial planet in the solar system and a primary focus of deep space exploration due to its potential habitability. Hydrated minerals, formed through water-rock interactions, provide essential insights into Mars’ early aqueous environment, geological evolution, and habitability. Hyperspectral remote sensing, with its ultra-high spectral resolution, has proven invaluable for identifying and quantifying these minerals. However, the sparse distribution and low abundance of hydrated minerals, along with challenges from spectral mixing and noise, have constrained current detection methods. These approaches, primarily relying on spectral parameters and visual interpretation, struggle to meet the demands of large-scale hyperspectral data processing. Recent advances in machine learning for terrestrial hyperspectral remote sensing offer innovative approaches to Martian mineral mapping, yet their application remains at an early stage. This review summarizes progress in the hyperspectral detection of Martian hydrated minerals, covering qualitative identification and quantitative abundance retrieval. It assesses the advantages, limitations, and applicability of existing methods and proposes future directions to advance this field.