Advances and challenges in plastic detection: a critical review of studies employing hyperspectral data

Plastic pollution is a growing global problem with severe environmental impacts. Hyperspectral remote sensing offers promising capabilities to detect plastics through their distinctive spectral signatures, enabling monitoring across different environments. Despite rapid technological advances and increasing research activity in this field, existing reviews rarely integrate hyperspectral laboratory spectra with image-based hyperspectral data from unmanned aerial vehicles, airborne sensors, and satellite platforms for plastic detection across diverse environments. This review is based on a structured literature search of the Web of Science, identifying studies published between 2014 and 2025; twenty-nine met the inclusion criteria and were examined to summarize the plastics investigated, the platforms and sensors used, data characteristics and analytical approaches. Key advances include the creation of spectral libraries, identification of diagnostic wavelengths in the shortwave infrared region and improvements in sensors and classification methods. Studies using unmanned aerial vehicles remain rare and often lack shortwave infrared coverage. Satellite observations provide wide geographical coverage but are limited in spatial and spectral resolution. Plastics such as polyethylene, polypropylene and polystyrene tend to exhibit more distinct spectral features and are more easily detected compared with polyethylene terephthalate, polyvinyl chloride or transparent plastics, particularly when wet or degraded. Detection performance also declines in the presence of environmental interference, mixed pixels and insufficient atmospheric or geometric correction. Future research should prioritize standardized datasets, integration of laboratory and image-based data using simulated and unmanned aerial vehicle platforms with full visible to shortwave infrared capability, and development of classification models validated under real environmental conditions.