I've noticed that ATM machines have become very good at reading handwriting on checks. I would like to write a program (using some appropriate machine learning or computer vision library) that is able to extract the following handwritten information from checks:
- Check amount
- Name (from the signature)
- Who the check is made out to
- Note or memo people write at the bottom of the check
I realize that reading sloppy handwriting is difficult, but it seems that with recent advances in machine learning computers are reaching human level ability to do this kind of task. If the algorithm only succeeds 75% of the time, that is still very useful, as long as the algorithm can also output some confidence score which indicates which checks will need human attention.
Question: Can you recommend machine learning or computer vision libraries which would be most useful for achieving this task? I've spent some time googling but have not found a clear answer.