Turkish lira banknotes classification using deep convolutional neural networks
Gulcin Baykal, Ugur Demir, Ira Shyti, Gozde Deniz Unal · 2018
While the technology improves rapidly in today's world, the fact that visually impaired people still face complications about monetary situations in their social life reveals technology is needed to propose a solution. In this study, a system to classify Turkish Lira banknotes is implemented with convolutional neural networks and the results of different architectures are compared. A new and unique dataset of Turkish Lira banknotes is prepared to train, evaluate and test the system. The state-of-the-art deep learning models are used with fine-tuning and as a result of comparison it is shown that DenseNet-121 architecture has achived 93,15% test accuracy on this dataset which is the best performance.