Blind people can have a really hard time recognizing paper currency while they can easily recognize coins due to their different texture, shape and thickness. In this repository, I try to solve the problem using Image Classification. There are ten different classes denoting all the paper currency denominations.
There are not a lot of images of Indian currency on the internet without watermarks. So I decided to create my dataset using different angles, zoom, noise, lighting condition. There are 55 images for each class for training(55 X 10 = 550 images) and 20 images for validation (20 X 10 = 200 images) New 20 rupees paper denomination is not included as it wasn't available with me
- MATLAB r2020b
- Deep Network Designer App
- Image Processing Toolbox