Lightweight CNN–Transformer and Explainable AI for Real-Time Counterfeit Currency Detection and Classification
IOSR Journal of Computer Engineering · 2026-05 · 28 (5) · 26-31
DOI: 10.9790/0661-2805012631 ↗Kurz gefasst
Fachartikel in der Zeitschrift IOSR Journal of Computer Engineering, 2026-05. Verfasst von Kumud Wasnik und 2 weiteren Autor:innen.
Abstract (Originalsprache)
This paper proposes a deployment-oriented artificial intelligence framework for real-time counterfeit currency authentication and denomination classification using hyperspectral imaging. Building on spectral-spatial representations, the proposed system combines informative spectral-band selection, a lightweight threedimensional convolutional stem, spectral-spatial attention, and a compact transformer encoder. The architecture is formulated as a multi-task network with authentication and denomination heads and an optional suspicious-region localization branch. Explainable AI is incorporated through Grad-CAM and attention-based visualizations so that the system can indicate regions contributing to a decision. To support practical deployment, model compression, quantization, knowledge distillation, and inference profiling are incorporated into the experimental design. The evaluation protocol measures not only classification quality but also latency, throughput, memory, parameter count, and model size. No experimental results are fabricated in this manuscript; the reported-result tables are designed to be completed from controlled experiments. The contribution is a reproducible blueprint for an edge-deployable HSI currency inspection system that jointly addresses authentication, classification, interpretability, and real-time constraints.
Zitieren
APA
Wasnik, K., Temurnikar, D. A., & Gadepaka, D. L. (2026). Lightweight CNN–Transformer and Explainable AI for Real-Time Counterfeit Currency Detection and Classification. IOSR Journal of Computer Engineering, 28(5), 26-31. https://doi.org/10.9790/0661-2805012631
Vancouver
Wasnik K, Temurnikar DA, Gadepaka DL. Lightweight CNN–Transformer and Explainable AI for Real-Time Counterfeit Currency Detection and Classification. IOSR Journal of Computer Engineering. 2026;28(5):26-31. doi:10.9790/0661-2805012631
BibTeX
@article{Wasnik2026,
author = {Wasnik, Kumud and Temurnikar, Dr. Ankit and Gadepaka, Dr. Latha},
title = {Lightweight CNN–Transformer and Explainable AI for Real-Time Counterfeit Currency Detection and Classification},
journal = {IOSR Journal of Computer Engineering},
year = {2026},
volume = {28},
number = {5},
pages = {26--31},
doi = {10.9790/0661-2805012631}
}
EUVIMED