The data and findings discussed here come from the cited paper and do not represent test results from VISPEK equipment.
The difference between handheld plastic identification devices depends not only on the nominal wavelength range, but also on the actual channel and response. This study allows different hardware to face the same type of recycling classification problems and provides a basis for comparison. When citing results, device-specific branches need to be retained to avoid transferring results from high-resolution instruments to discrete-channel devices.
Summary
Research on multi-position scanning of common resin objects and compare the classification performance of continuous near-infrared and discrete channel equipment. This case corresponds to the SpectraPod independent branch and cannot use high-resolution NIRS results. The results are affected by color, coating and multi-layer structure, and it is not clear whether repeated records are isolated by object; the unknown label does not mean that any unknown plastic can be identified.
(AI-generated conceptual illustration for context; it is not data from the paper, a photograph of the research equipment, or a VISPEK product diagram.)
Why It Matters
Low-pass plastic sorting has been compared with independent devices, providing a direct reference for the development of portable material identification at recycling sites.
Methods and Reference Measurements
The study included 198 pieces of plastic: PET44, HDPE39, PVC22, LDPE13, PP56, and PS24. The categories were determined according to the resin identification; multiple parts and orientations were scanned, and the three devices totaled 2,406 records. A high-resolution near-infrared instrument and two discrete-channel devices collect reflection data, add adjacent wavelength differences, and compare SVM, XGBoost, random forest, and Gaussian Naive Bayes. Hierarchical nesting verification using 5-fold outer layer and 10-fold inner layer parameter adjustment is adopted; whether repeated records of the same object are grouped and isolated is not clearly reported, and the optical path is not reported.
Wavelengths and Acquisition Parameters
| Band/parameter | Original meaning and usage instructions |
|---|---|
| 1050-1950 nm | High-resolution NIRS independent control range. |
| 850-1600 nm | 16-channel total response range discussed on page 9 of SpectraPod text; discrete overlapping response, not continuous spectrum. |
| 900-1800 nm LEDs; 900-1700 nm detector response | The Plastic Scanner has an independent branch, and the light source and detector ranges cannot be confused. |
Key Findings
Nested cross-validation classification mean: high-resolution NIRS 0.97±0.01, SpectraPod 0.93±0.01, Plastic Scanner 0.70±0.03. The preferred applications in this article are only for SpectraPod branches and cannot borrow NIRS results.
Limitations and Interpretation
Color, coatings, multilayer packaging, and low-reflective samples can alter response; unknown class labels are not proven to identify any unknown plastic.
Band and model boundaries: Only SpectraPod multi-channel independent classification is included, the text discusses the total response 850-1600 nm; the 1050-1950 nm high-resolution instrument 0.97 result is not used instead of SpectraPod0.93. The response curve and channel overlap must be checked separately, and it is not called a continuous spectrum.
Image copyright belongs to the paper’s original authors. License: CC BY 4.0.
Source Paper
van Hoorn H, Pourmohammadi F, de Leeuw AW, Vasulkar A, de Vos J, van den Berg S. (2025). Machine Learning-Based Identification of Plastic Types Using Handheld Spectrometers. Sensors (Basel, Switzerland). https://doi.org/10.3390/s25123777
Contact Us
Whether portable equipment can be used for similar detection needs to be evaluated based on samples and experimental conditions, and the specific effect still needs to be verified through actual testing. Welcome to contact VISPEK to exchange your sample and testing needs, jointly evaluate experimental plans, and explore the feasibility of using our company's equipment for actual measurement and verification.
- Email: contact@vispek.cn
- Website: https://www.vispek.cn
VISPEK products shown for illustration; these were not the devices used in the cited study.