AI-Enhanced Terahertz Spectroscopy for Real-Time Classification of Plastics for Recycling and Sorting Applications

Rejeena R. Sebastian; Charles Simard; Redwan Ahmad; Leyla Agosti Cagalli Burri; Christian Tardif; Francois Blanchard

doi:10.1109/PN70808.2026.11657738

AI-Enhanced Terahertz Spectroscopy for Real-Time Classification of Plastics for Recycling and Sorting Applications

Rejeena R. Sebastian
, Charles Simard
, Redwan Ahmad
, Leyla Agosti Cagalli Burri
, Christian Tardif
, Francois Blanchard

Department of Electrical Engineering
École de technologie supérieure
Luqia

Résultats de recherche: Contribution à un journal › Article de conférence › Revue par des pairs

Résumé

We present a real-time terahertz (THz) plastic classification system based on rotating frequency-selective surfaces (FSSs) and supervised machine learning. The system achieves an average accuracy of 95% across seven polymer types, with prediction times under one second. The analysis confirms robust feature learning and reliable discrimination of chemically similar polymers, demonstrating strong potential for practical recycling and sorting applications.

langue originale	Anglais
journal	Photonics North, PN
Numéro de publication	2026
Les DOIs	https://doi.org/10.1109/PN70808.2026.11657738
état	Publié - 2026
Evénement	2026 Photonics North, PN 2026 - Quebec City, Canada Durée: 2 juin 2026 → 5 juin 2026

Accès au document

10.1109/PN70808.2026.11657738

Autres fichiers et liens

Lien vers la publication dans Scopus

Empreinte digitale

Voici les principaux termes ou expressions associés à « AI-Enhanced Terahertz Spectroscopy for Real-Time Classification of Plastics for Recycling and Sorting Applications ». Ces libellés thématiques sont générés à partir du titre et du résumé de la publication. Ensemble, ils forment une empreinte digitale unique.

Citer cette ressource

@article{abcfd593651e49f5bfb3c8f3a7b01dc7,

title = "AI-Enhanced Terahertz Spectroscopy for Real-Time Classification of Plastics for Recycling and Sorting Applications",

abstract = "We present a real-time terahertz (THz) plastic classification system based on rotating frequency-selective surfaces (FSSs) and supervised machine learning. The system achieves an average accuracy of 95\% across seven polymer types, with prediction times under one second. The analysis confirms robust feature learning and reliable discrimination of chemically similar polymers, demonstrating strong potential for practical recycling and sorting applications.",

keywords = "multispectral spectroscopy, plastic recycling, real-time sensing, supervised learning, Terahertz",

author = "Sebastian, \{Rejeena R.\} and Charles Simard and Redwan Ahmad and Burri, \{Leyla Agosti Cagalli\} and Christian Tardif and Francois Blanchard",

note = "Publisher Copyright: {\textcopyright} 2026 IEEE.; 2026 Photonics North, PN 2026 ; Conference date: 02-06-2026 Through 05-06-2026",

year = "2026",

doi = "10.1109/PN70808.2026.11657738",

language = "English",

journal = "Photonics North, PN",

issn = "2693-8324",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

number = "2026",

}

TY - JOUR

T1 - AI-Enhanced Terahertz Spectroscopy for Real-Time Classification of Plastics for Recycling and Sorting Applications

AU - Sebastian, Rejeena R.

AU - Simard, Charles

AU - Ahmad, Redwan

AU - Burri, Leyla Agosti Cagalli

AU - Tardif, Christian

AU - Blanchard, Francois

PY - 2026

Y1 - 2026

N2 - We present a real-time terahertz (THz) plastic classification system based on rotating frequency-selective surfaces (FSSs) and supervised machine learning. The system achieves an average accuracy of 95% across seven polymer types, with prediction times under one second. The analysis confirms robust feature learning and reliable discrimination of chemically similar polymers, demonstrating strong potential for practical recycling and sorting applications.

AB - We present a real-time terahertz (THz) plastic classification system based on rotating frequency-selective surfaces (FSSs) and supervised machine learning. The system achieves an average accuracy of 95% across seven polymer types, with prediction times under one second. The analysis confirms robust feature learning and reliable discrimination of chemically similar polymers, demonstrating strong potential for practical recycling and sorting applications.

KW - multispectral spectroscopy

KW - plastic recycling

KW - real-time sensing

KW - supervised learning

KW - Terahertz

UR - https://www.scopus.com/pages/publications/105049052807

U2 - 10.1109/PN70808.2026.11657738

DO - 10.1109/PN70808.2026.11657738

M3 - Conference article

AN - SCOPUS:105049052807

SN - 2693-8324

JO - Photonics North, PN

JF - Photonics North, PN

IS - 2026

T2 - 2026 Photonics North, PN 2026

Y2 - 2 June 2026 through 5 June 2026

ER -