A Deep Learning Approach for Detecting Bangladeshi Counterfeit Currency

Abstract

Fake or duplicate currency made with the intent to deceive is referred to as counterfeit money. According to recent sources, demonetization resulted in an all-time high inflow of bogus notes into banks, increasing questionable activities. The method of printing or supplying counterfeit currency is growing more modern day by day as the era progresses. In addition, the detection of counterfeit currency is being upgraded. Currency detection has already attracted a lot of attention. In this day and age of information technology, counterfeit currency may be identified using a variety of tools, gadgets, and software. Most current efforts to detect a counterfeit note rely on image processing techniques. Many of the pieces have received positive feedback, but some have flaws. The number of research which has worked with all the Bangladeshi banknotes is very insignificant. That is why we are focusing on this issue. In this method, we have been working with Bangladeshi 5 taka, 10 taka, 20 taka, 50 taka, 100 taka, 200 taka, 500 taka, 1000 taka. We created a dataset for Bangladeshi currency named BANGLADESHI BANKNOTE which contains 8000 images. In this research Modified AlexNet (M-AlexNet) is used for feature extraction from BANGLADESHI BANKNOTE where classification is followed by Multi Support Vector Machine (M-SVM).