I'd like to create a MRZ detection system using a CNN : inputs are images of contracts and outputs are zones where to read. All the contracts have the same format but pictures can differ (light, angle...I think about applying an OCR on cropped images of zones predicted).

I'm wondering about the structure of NN, I think about creating 02*03 output neurons (supposing there are three outputs : name, date, location) for predicting top left and bottom right coordinates of each zone.

Is it good idea? Can it scale if the number of parameters grows (13)?



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy