Affordable Smart Vending Machine with Deep Learning-Based Real-Time Bangladeshi Currency Recognition
Hakim Mohammad Insaf, Md. Atikur Rahman, Md. Zahidur Rahman, Ohidujjaman Ohidujjaman, Mahmudul Hasan · 2025
This paper presents the design and development of a cost-effective and multipurpose vending machine that is suitable for the Bangladeshi market. The proposed system integrates three different payment options: coin, paper currency, and online payment. A custom CNN model is trained to identify and recognize Bangladeshi currency notes, achieving an accuracy of 93 %. The purpose of the research is to develop this machine using our own technology and make it cost-effective by utilizing local components and open-source technologies like Arduino. This approach will allow the technology to be used in ATM booths, automatic ticket machines in railway or metro stations, and many other applications across the country. This research has the potential to significantly reduce the cost associated with purchasing such expensive technology.