I know how to train a NN for recognizing handwritten digits (e.g. using the MNIST database).

I'm wondering how to accomplish the same "online", which is during the process of writing e.g. I'm writing down a digit, and during the process and while keeping "the pen down", the NN recognizes the digit. To make it simpler, I could assume one stroke-order only for each digit (e.g. each digit is written with a certain order of strokes, like for digit 1 is just a vertical line going up to down).

Which is the most suitable NN for the purpose and how to accomplish this?

  • $\begingroup$ Have you tried to use a neural network trained on MNIST to achieve that? You can pass the partial input the this NN and observe the output of it and how it evolves. $\endgroup$ – nbro Sep 15 '19 at 11:57
  • 1
    $\begingroup$ @nbro I haven't tried. However it would surely be sub-optimal, as a static partial input lacks "temporal" information (e.g. direction of stroke). $\endgroup$ – kuma Sep 15 '19 at 15:41
  • $\begingroup$ This is a good point! $\endgroup$ – nbro Sep 15 '19 at 15:51

Your Answer

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

Browse other questions tagged or ask your own question.