I am looking for a person with knowledge of machine learning, who will undertake to write a mobile application that will check the originality of the document, which will implement three functionalities:
1. After hovering the phone camera over the banknote, the application recognizes the denomination of the banknote and displays the information that it is, for example, PLN 200.
We limit ourselves here only to Polish denominations of 10, 20, 50, 100, 200 and 500 zlotys.
Here, the application should be trained to recognize the denominations of banknotes on the basis of the created dataset.
The dataset should contain examples of right and left banknote denominations.
2. Based on the fast Fourier FFT transform, the application can verify the originality of the banknote after taking a photo of the banknote. Google: FFT in images.
Color photo -> preprocessing (area cut, resolution adjustment etc) -> black and white conversion -> FFT -> machine learning.
Such an area on the banknote may be the banknote laterite bar on the right, front or left back side of the banknote.
3. Reading banknote numbering and saving it to another txt or excel file. (OCR)
The application needed as soon as possible.
The proposed price for the work is PLN 2,200.