Fake Indian Currency Recognition Using CNN and Mobile Net Algorithm
Keywords:
Counterfeit Detection, Convolutional Neural Networks, MobileNet, Indian Currency, Image Classification, Machine LearningAbstract
The rise of counterfeit currency is a serious threat to our economic stability, which is why we need smarter ways to detect it. This project, called "Identification of Fake Indian Currency Using Convolutional Neural Network and MobileNet Algorithm," introduces an innovative method for spotting counterfeit notes by using deep learning techniques. The research focuses on the MobileNet model to differentiate between real Indian currencies and fakes. MobileNet is well-regarded for its efficiency and accuracy in image classification, and here, we put it to the test for identifying counterfeit money. The model is fine-tuned for both speed and precision, making it a great fit for real-time applications in counterfeit detection systems.
Downloads
References
Howard et al. (2017) introduced MobileNets, a lightweight CNN architecture optimized for mobile and embedded vision tasks using depthwise separable convolutions [arXiv:1704.04861].
Zhang et al. (2018) proposed ShuffleNet, a compact neural network that incorporates pointwise group convolutions and channel shuffling to enhance computational efficiency on mobile devices (CVPR).
He et al. (2016) developed ResNet, a deep residual learning framework that significantly improved image recognition performance by enabling deeper network architectures (CVPR).
Tan and Le (2019) presented EfficientNet, which introduces a compound scaling method for uniformly scaling CNN dimensions, achieving state-of-the-art accuracy with fewer parameters (ICML).
Khoda et al. (2020) designed a hybrid model combining CNN with Random Forest for counterfeit currency detection, leveraging image-based features for classification (JEEECS, 29(1), 45–52).
Szegedy et al. (2017) extended the Inception family with Inception-v4 and integrated residual connections to further boost network learning and convergence (AAAI Conf. on Artificial Intelligence, Vol. 31, No. 1).
Chollet (2017) introduced Xception, which improves upon Inception by using depthwise separable convolutions throughout the network to enhance performance and reduce complexity (CVPR).
Krizhevsky, Sutskever, and Hinton (2012) developed AlexNet, a pioneering deep CNN model that won the ImageNet challenge and catalyzed deep learning in computer vision (NeurIPS 25).
Liu et al. (2016) proposed SSD (Single Shot MultiBox Detector) for real-time object detection, capable of detecting multiple object categories in a single forward pass (ECCV).
Simonyan and Zisserman (2014) introduced VGGNet, a deep convolutional network that demonstrated the benefits of depth in improving large-scale image recognition performance [arXiv: 1409.1556].
Downloads
Published
Issue
Section
License
Copyright (c) 2025 International Journal of Scientific Research in Computer Science, Engineering and Information Technology
This work is licensed under a Creative Commons Attribution 4.0 International License.