We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
Research on Waste Plastics Classification Method Based on Multi-Scale Feature Fusion
Summary
Researchers developed a multi-scale feature fusion method combining RGB and hyperspectral near-infrared imaging with a SOLOv1 instance segmentation framework to classify transparent PET, colored PET, and PP plastic bottles on a conveyor belt, achieving improved identification accuracy for transparent materials that conventional single-sensor methods struggle to detect.
Microplastic particles produced by non-degradable waste plastic bottles have a critical impact on the environment. Reasonable recycling is a premise that protects the environment and improves economic benefits. In this paper, a multi-scale feature fusion method for RGB and hyperspectral images based on Segmenting Objects by Locations (RHFF-SOLOv1) is proposed, which uses multi-sensor fusion technology to improve the accuracy of identifying transparent polyethylene terephthalate (PET) bottles, blue PET bottles, and transparent polypropylene (PP) bottles on a black conveyor belt. A line-scan camera and near-infrared (NIR) hyperspectral camera covering the spectral range from 935.9 nm to 1722.5 nm are used to obtain RGB and hyperspectral images synchronously. Moreover, we propose a hyperspectral feature band selection method that effectively reduces the dimensionality and selects the bands from 1087.6 nm to 1285.1 nm as the features of the hyperspectral image. The results show that the proposed fusion method improves the accuracy of plastic bottle classification compared with the SOLOv1 method, and the overall accuracy is 95.55%. Finally, compared with other space-spectral fusion methods, RHFF-SOLOv1 is superior to most of them and achieves the best (97.5%) accuracy in blue bottle classification.
Sign in to start a discussion.
More Papers Like This
Advancing Plastic Waste Classification and Recycling Efficiency: Integrating Image Sensors and Deep Learning Algorithms
Researchers developed a deep learning approach combined with image sensors to improve plastic waste classification and recycling efficiency. The study demonstrates that this method can distinguish between chemically similar plastics like PET and PET-G that conventional near-infrared spectroscopy struggles to differentiate, potentially improving automated sorting systems.
Rapid and Nondestructive On-Site Classification Method for Consumer-Grade Plastics Based on Portable NIR Spectrometer and Machine Learning
Researchers used a portable near-infrared spectrometer combined with machine learning to rapidly identify and classify seven types of consumer plastic waste on-site without damaging the samples. Faster and cheaper plastic identification tools are important for improving plastic recycling efficiency and ultimately reducing the amount of plastic that ends up as microplastic pollution.
SpectraNet: A unified deep learning framework for infrared spectroscopy-based prediction of plastic recyclability, type classification, and microplastic identification
Researchers built SpectraNet, a deep learning framework using mid-infrared spectroscopy to perform three tasks—plastic recyclability assessment, polymer type classification, and microplastic identification—supported by an open-access infrared spectral database of plastics and microplastics.
Comparative Study of Chemometric Approaches and Machine Learning for Miniaturized Near-infrared (micronir) Spectroscopy in Plasticwaste Sorting
This study tested a miniaturized near-infrared (NIR) spectroscopy device combined with chemometric and machine learning methods to sort different types of plastic waste. The approach accurately identified polymer types, supporting more efficient plastic recycling operations that could reduce microplastic generation.
Rapid detection of colored and colorless macro- and micro-plastics in complex environment via near-infrared spectroscopy and machine learning.
Researchers developed a near-infrared spectroscopy method combined with machine learning classifiers -- including PLS-DA, random forest, and XGBoost -- to rapidly identify both colored and colorless plastic fragments across different polymer types, thicknesses, and environmental backgrounds. The approach improved detection of colorless plastics that are typically underestimated in environmental surveys, with random forest achieving the highest classification accuracy.