Jordanian Currency Recognition Using Deep Learning

Automatic Currency Recognition (ACR) has a significant role in various domains, such as assessment of visually impaired people, banking transactions, counterfeit detection, digital transformation, currency exchange, vendor machines, etc. Therefore, developing an accurate ACR system enhances efficiency across several domains. The contribution of this paper is three-fold; it proposed a large dataset of 2799 images and seven denominations for Jordanian currency recognition. The second contribution proposed an efficient multiscale VGG net to recognize Jordanian currency. Third, popular CNN architectures on the proposed dataset will be evaluated, and the result will be compared with the proposed architectures. Four metrics were used in the evaluation. The experimental result showed the accuracy of the proposed Multiscale VGG outperformed VGG16, DenseNet121, ResNet50, and ResNet101 and achieved 99.88%, 99.88%, 99.89%, and 99.98% accuracy, precision, sensitivity, and specificity.

@article{Alghyaline2025,
  title     = {Jordanian Currency Recognition Using Deep Learning},
  journal   = {International Journal of Advanced Computer Science and Applications},
  volume    = {16},
  number    = {1},
  year      = {2025},
  publisher = {The Science and Information Organization},
  author    = {Salah Alghyaline},
  doi       = {10.14569/IJACSA.2025.0160139},
  url       = {https://doi.org/10.14569/IJACSA.2025.0160139}
}