Plastic pellets (2-5 mm industrial feedstock granules) are an increasingly important component of plastic pollution in aquatic environments. Their small size, variable weathering state, and similarity to natural substrates such as sand and gravel make their detection and characterization challenging. This study, conducted within the SSPIRIT project, investigates a complementary remote sensing approach for detecting and mapping plastic pellets on riverbanks and coastal environments.<br />Two detection strategies are explored. First, very high spatial resolution RGB imagery acquired by drones and smartphones is used to detect individual pellets and estimate their abundance and spatial distribution. Second, SWIR hyperspectral imaging is employed to exploit the distinctive spectral signatures of common polymers such as polyethylene, polypropylene, polystyrene, and PVC. This enables both discrimination of pellets from natural background materials and identification of dominant polymer types.<br />A dedicated spectral library is being developed by combining existing open-access measurements with new laboratory spectra of virgin and weathered pellets, including bio fouled samples. Field acquisitions using a VNIR-SWIR hyperspectral imaging system will support the evaluation of sparse spectral unmixing techniques for estimating pellet abundances within mixed pixels. Initial RGB and multispectral datasets have already been acquired, while spectral measurements and field campaigns are ongoing. Preliminary results will be presented, together with an assessment of how high-resolution feature detection and hyperspectral unmixing can be combined to improve plastic pellet monitoring in riverine and coastal environments.