OCR Cheque Scanner for a Banking App

computer vision


An international bank. They were working on a technology to simplify and automate cheque transactions for their customers and enhance their banking app.


The client tried several solutions, including Open-Source Software. It could not deliver accurate results and precisely scan handwritten numbers.

They approached Broutonlab Data Science Consulting to develop a model for scanning handwritten cheques.


Automated reading and verification of cheques to save costs and increase workforce efficiency.

Data Scientists at BroutonLab trained a neural network to process the input in the form of an image and return it as numbers. We developed a highly precise analyzer with an accuracy rate of 99%.

It is an end2end data-driven solution, which doesn’t contain any algorithmic heuristics and rules.

As the number of labeled images increases, the engine becomes more precise. The bank can continue enhancing the accuracy of the model with no prior knowledge of deep learning or programming.

Talk To Us Now