This system uses Machine Learning to classify bank notes. Trained on self-taken images of up to 7,000 taken under different various conditions of the environment and money conditions such as wet, crunched and dirty notes. it currently has only been trained and tested on Malawian bank notes. The system is continuously trained and tested for improvements.

Using computer vision, it gets an image of a bank note and starts to look for the specific features that make that money unique. it then cleans the image data and feeds it to the frozen convolution neural network model that spits out a result. The result is an array of labels which in our case are names of the various bank notes we've trained on and the confidence levels for each. The system will pick out the label with the highest confidence level.