I'm on a project that process hundreds of forms. The forms have consistent formats but are filled out by hand by different people. I need a way to quickly processing all of these data into electronic forms. OCR recognition for typed document seems mature but for hand-writting is very lacking.

For this thought, let's consider a form with several fields like this. Field_1: Value1 (example, Name: John, where Name is Field and John is value)

Considering that forms are structured and typed, OCR should be able to recognize the fields. However, for the values of the fields, they are written and OCR will perform very poorly.

So is there a way where the fields would be recognized on the imagem, then a image chunk of the value would be returned?

Thanks.