A recent study conducted by Shanghai researchers developed a high-accuracy model for identifying plastic pollution.
In a recent study, a research collaboration from Tongji University and the Shanghai Institute of Pollution Control and Ecological Security explored a new method for detecting plastic pollution in the environment. This novel method, which used near-infrared hyperspectral imaging (NIR-HSI) and advanced machine learning (ML) models, was designed to detect colorless fragments of microplastics, which have been overlooked by other detection methods. The findings of this study were in the Journal of Environmental Sciences (1).
Microplastics are small particles of disposed plastic that usually are 5 mm or smaller (2). Recently, there has been an increased focus to monitor and mitigate plastic waste in the environment (3). However, colorless fragments of plastic have one mostly unnoticed, and most current detection methods cannot pick up these bits of plastic.
Close up side shot of microplastics lay on people hand. Concept of water pollution and global warming. Climate change idea. | Image Credit: © Pcess609 - stock.adobe.com
In this study, the researchers explored using a new technique that was highly accurate, non-destructive, and able to conduct in situ detection. The research team used NIR spectroscopy to classify and identify both colored and colorless plastic fragments in complex environmental backgrounds (1). Four ML models were tested: partial least squares–discriminant analysis (PLS-DA), extreme gradient boost (XGBOOST), support vector machine (SVM), and random forest classifier (RFC). In doing so, the researchers sought to determine their efficiency in identifying plastic particles of various colors, thicknesses, and polymer types (1).
Out of the four models, PLS-DA performed the best. However, all models exhibited challenges in correctly classifying colorless plastic fragments—particularly when their thickness was below 0.1 mm—as they were often misinterpreted as part of the background (1).
A two-stage classification method was proven to be helpful in solving this challenge. Because colorless fragments are often ignored as background material, the approach first distinguishes plastic types and then re-examine it. The result was an accuracy of 99% in detecting the colorless fragments over different environmental settings (1).
Another noteworthy aspect to this study was how the method had a high-throughput detection capability. By integrating NIR-based hyperspectral imaging (HSI) with ML, Zhang’s team successfully created a rapid, non-invasive technique that can identify 15 types of macroplastics, microplastics, and rubber particles within various organic and inorganic environmental backgrounds—without requiring labor-intensive sample preparation (1).
The ability to detect both colored and colorless plastics with high precision is a major advancement in plastic pollution research. Traditional microplastic identification methods often rely on tedious extraction and purification steps, limiting their scalability (1). By eliminating these bottlenecks, this new technique paves the way for large-scale environmental monitoring programs, enabling faster and more comprehensive assessments of plastic contamination (1).
The findings in this study show that colorless plastic can be detected with high accuracy. The developed method prevents the underestimation of plastic pollution, ensuring that colorless microplastics are accurately accounted for in environmental surveys (1).
Moreover, the study’s approach could be adapted for industrial applications, such as waste sorting and recycling processes, where distinguishing different types of plastics quickly and efficiently is essential (1).
By leveraging the power of NIR hyperspectral imaging and machine learning, the new methodology developed by Zhang and his colleagues could help improve the accuracy of colorless microplastic identification. With plastic pollution continuing to threaten marine and terrestrial ecosystems, such technological advancements will play a crucial role in shaping future environmental policies and sustainability initiatives (1). This research marks a critical step toward more efficient, large-scale monitoring of plastic waste, providing a promising solution for tackling one of the world’s most persistent environmental challenges (1).
Radar and Soil Spectroscopy Boost Soil Carbon Predictions in Brazil’s Semi-Arid Regions
July 7th 2025A new study published in Geoderma demonstrates that combining soil spectroscopy with radar-derived vegetation indices and environmental data significantly improves the accuracy of soil organic carbon predictions in Brazil’s semi-arid regions.
Deciphering Fossil Preservation in Brazil’s São Carlos Shale Using Analytical Spectroscopy
July 7th 2025A new study published in the Journal of South American Earth Sciences reveals how microbial activity, low pH conditions, and sediment chemistry in Brazil’s São Carlos Shale uniquely preserved diverse Upper Cretaceous fossils, offering fresh insights into the paleoenvironment of the Bauru Basin.
Evaluating Microplastic Detection with Fluorescence Microscopy and Raman Spectroscopy
July 2nd 2025A recent study presented a dual-method approach combining confocal micro-Raman spectroscopy and Nile Red-assisted fluorescence microscopy to enhance the accuracy and throughput of microplastics detection in environmental samples.
AI and Dual-Sensor Spectroscopy Supercharge Antibiotic Fermentation
June 30th 2025Researchers from Chinese universities have developed an AI-powered platform that combines near-infrared (NIR) and Raman spectroscopy for real-time monitoring and control of antibiotic production, boosting efficiency by over 30%.
Toward a Generalizable Model of Diffuse Reflectance in Particulate Systems
June 30th 2025This tutorial examines the modeling of diffuse reflectance (DR) in complex particulate samples, such as powders and granular solids. Traditional theoretical frameworks like empirical absorbance, Kubelka-Munk, radiative transfer theory (RTT), and the Hapke model are presented in standard and matrix notation where applicable. Their advantages and limitations are highlighted, particularly for heterogeneous particle size distributions and real-world variations in the optical properties of particulate samples. Hybrid and emerging computational strategies, including Monte Carlo methods, full-wave numerical solvers, and machine learning (ML) models, are evaluated for their potential to produce more generalizable prediction models.