Towards Low-cost Plastic Recognition using Machine Learning and Multi-spectra Near-infrared Sensor

Aug 3, 2026

A low-cost multi-spectral near-infrared sensor combined with machine learning enables plastic recognition for daily-used objects and waste sorting. The sensor measures 64 wavelength bands, while four computational models—Random Forest, Support Vector Machines, Multi-Layer Perceptron, and Convolutional Neural Networks—process the spectral data. Experiments use samples from six household plastic waste types plus virgin plastics for comparison. Results highlight best recognition performance by CNN and SVM, supporting affordable portable applications in recycling, healthcare, agriculture, e-waste, and manufacturing.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages5

Reading Time13 min

AccessFree to download

Introduction

Low-cost plastic recognition approach

Multi-spectral NIR sensor and ML pipeline

Data collection and validation

Citation for published version: Martinez Hernandez, U, West, G & Assaf, T 2023, Towards Low-cost Plastic Recognition using Machine Learning and Multi-spectra Near-infrared Sensor. in 2023 IEEE SENSORS. IEEE, IEEE Sensors 2023, Vienna, Austria, 29/10/23 . [https://doi.org/10.1109/SENSORS56945.2023.10325140](https://doi.org/10.1109/SENSORS56945.2023.10325140) DOI: 10.1109/SENSORS56945.2023.10325140 Publication date: 2023 Document Version Peer reviewed version Link to publication Publisher Rights CC BY University of Bath Alternative formats If you require this document in an alternative format, please contact: [openaccess@bath.ac.uk](openaccess@bath.ac.uk) General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. Take down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim. Download date: 17. Sep. 2025 Towards Low-cost Plastic Recognition using Machine Learning and Multi-spectra Near-infrared Sensor Gregory West, Tareq Assaf, Uriel Martinez-Hernandez* Abstract—This work presents a low-cost sensor and machine learning methods approach for plastic recognition in daily used objects. The sensor is a multi-spectral near-infrared sensor capable of measuring 64 wavelength. Data processing and analysis are performed using a set of four machine learning based computational methods (Random Forest, Support Vector Machines, Multi-Layer Perceptron, Convolutional Neural Networks). Validation is performed by collecting data samples from 6 different types of waste plastics found in household recycling and virgin materials. The results show that Convolutional Neural Networks and Support Vector Machines achieve the highest recognition accuracy of 62.08% with waste plastics and 54.72% with virgin plastics, respectively. The results show how this low-cost multi-spectral near-infrared sensor and machine learning can be effective in plastic recognition tasks and potentially enables to create new applications in other fields that require affordable and portable solutions such as in agriculture, e-waste recycling, healthcare and manufacturing. I. INTRODUCTION Plastics are versatile, durable, lightweight and low-cost materials essential for the world economy with the manufacturing of daily usage products such as bottles, food containers, bags, telephones [1], [2] . Household waste and daily usage products typically contain one of the following types of plastics: Polyethylene Terephthalate (PET), Polypropylene (PP), Polystyrene (PS), Polyvinyl Chloride (PVC), LowDensity Polyethylene (LDPE), High-Density Polyethylene (HDPE) . Even though the key role played by plastics, their uncontrolled used in mass and low biodegradation process have contributed to the generation of millions of tons of waste, contamination of the environment, increasing carbon levels and health problems [3], [4], [5] . Therefore, it is important to develop intelligent and affordable systems that can recognise plastic types for posterior sorting and recycling reducing the generation of more plastics [6], [7], [8] . Automated systems for plastic waste identification commonly use different sensing technologies to extract object properties such as X-ray fluorescence (XRF), Raman spectroscopy, near-infrared spectroscopy (NIR), miniaturised near-infrared (MicroNIR) and laser-induced breakdown spectroscopy (LIBS) [9],[10],[11] . Data from these sensing technologies has been used together with computational methods including Linear Discriminant Analysis (LDA), Support Vector Machines (SVM), Principal Component Analysis (PCA), *[Corresponding. author:](Corresponding. author:) Uriel Martinez-Hern

FAQs

What sensor approach is used for plastic recognition?

The work uses a low-cost multi-spectral near-infrared sensor capable of measuring 64 wavelength values (AS7241) to extract properties from different plastics.

Which machine learning models are evaluated?

Random Forest, Support Vector Machines, Multi-Layer Perceptron, and Convolutional Neural Networks are evaluated on the spectral data.

How is validation performed and what accuracies are reported?

Validation is done using data from six types of household waste plastics and comparing against virgin plastics. CNN and SVM achieve the highest recognition accuracy, reaching 62.08% for waste plastics and 54.72% for virgin plastics, respectively.

Aug 3, 2026

Upload by Evangeline