Asian Currency Identification and Recognition Using Lightweight Convolutional Neural Network

Currency identification and recognition system has been a useful tool in numerous occasions in today's rapid growing of technology. Currency identification plays a vital role in financial applications such as ATMs, vending machines, and mobile payment solutions. Asian currencies in particular tend to confuse people, such as tourist, of what kind of Asian banknote they hold. This study introduces an innovative approach in identifying and recognizing Asian currencies utilizing a lightweight convolutional neural network. This model is designed to accurately identify and classify the Asian currency and its denomination. The researchers used Raspberry Pi Model 4 B as the microcontroller for the system and implemented the Raspberry Pi Camera V2 with its 8-megapixel Sony IMX219 to acquire the images of Asian currencies. The training procedure of the algorithm for the model is carried out initially with the creation of custom datasets and processed through a computer. Based on the confusion matrix and the results presented, the overall accuracy of the model is 96.82% in detecting the banknotes.