Non-Targeted Electronic-Nose Fingerprinting and Microscopic Imaging for Leather Classification via Transformer-Based Multimodal Fusion
- 1College of Information Science and Technology, Beijing University of Chemical Technology, Beijing 100029, China
- 2School of Biomedical Engineering, Capital Medical University, Beijing 100069, China
Abstract
The rapid, non-destructive screening of leather authenticity requires analytical measurements that capture complementary volatile-response and microstructural information. This study developed RLFormer, a multimodal workflow combining cross-responsive metal-oxide-semiconductor electronic-nose (E-nose) fingerprints with microscopic surface images. Four classes—synthetic, pig, sheep, and cattle leather—were represented by 100 specimens each and measured during a standardized 10-day campaign. The relative standard deviations of the category-level daily mean E-nose responses ranged from 1.4 to 13.3% (global mean, 6.18%), and a variance-component analysis revealed a progressive between-day response attenuation common to all categories. A ResNet50 visual encoder and a two-layer long short-term memory (LSTM) encoder generated modality-specific representations that were integrated by a transformer encoder. With days 1–7 used for training and days 8–10 for testing, RLFormer achieved an accuracy of 97.67±2.16% and a Cohen’s kappa of 0.9689±0.0288 over five random seeds, compared with an 87.50±2.50% accuracy for direct ResNet50–LSTM fusion. The results demonstrate the value of combining non-targeted volatile-response patterns with microscopic texture for rapid leather classification. The workflow is intended for screening rather than compound-specific identification and requires independent validation before deployment.
Related
- Fingerprinting of volatile profiles of sprouted and unsprouted seeds and flours of Phaseolus vulgaris cultivars using HS-SPME/GC-MS
- Non-Invasive Trace Element Fingerprinting of the Lion Man, a Left Mammoth Tusk and Mammoth Ivory Fragments Found at the Hohlenstein-Stadel Cave, Swabian Alb, Germany
- Phytochemical Fingerprinting of Nigerian Versus South African Loranthus micranthus Linn. Extracts and Their Antimicrobial Potentials Against Five Resistant Bacteria
- From Dynamic Frustration to Steric Trapping: The Microscopic Mechanism of a Monotropic Phase Transition
- Microscopic Model of Exciton Polarons in 2D Wigner Crystals
BibTeX
@article{wang2026nontargeted,
title = {{Non-Targeted Electronic-Nose Fingerprinting and Microscopic Imaging for Leather Classification via Transformer-Based Multimodal Fusion}},
author = {Xiaoran Wang and Juan Chen and Yu Gu},
journal = {Molecules},
year = {2026},
month = oct,
doi = {10.3390/molecules31213879},
url = {https://www.mdpi.com/1420-3049/31/21/3879},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Non-Targeted Electronic-Nose Fingerprinting and Microscopic Imaging for Leather Classification via Transformer-Based Multimodal Fusion
AU - Xiaoran Wang
AU - Juan Chen
AU - Yu Gu
PY - 2026
DA - 2026/10/27
JO - Molecules
DO - 10.3390/molecules31213879
UR - https://www.mdpi.com/1420-3049/31/21/3879
AB - The rapid, non-destructive screening of leather authenticity requires analytical measurements that capture complementary volatile-response and microstructural information. This study developed RLFormer, a multimodal workflow combining cross-responsive metal-oxide-semiconductor electronic-nose (E-nose) fingerprints with microscopic surface images. Four classes—synthetic, pig, sheep, and cattle leather—were represented by 100 specimens each and measured during a standardized 10-day campaign. The relative standard deviations of the category-level daily mean E-nose responses ranged from 1.4 to 13.3% (global mean, 6.18%), and a variance-component analysis revealed a progressive between-day response attenuation common to all categories. A ResNet50 visual encoder and a two-layer long short-term memory (LSTM) encoder generated modality-specific representations that were integrated by a transformer encoder. With days 1–7 used for training and days 8–10 for testing, RLFormer achieved an accuracy of 97.67±2.16% and a Cohen’s kappa of 0.9689±0.0288 over five random seeds, compared with an 87.50±2.50% accuracy for direct ResNet50–LSTM fusion. The results demonstrate the value of combining non-targeted volatile-response patterns with microscopic texture for rapid leather classification. The workflow is intended for screening rather than compound-specific identification and requires independent validation before deployment.
SN - 1420-3049
ER -