If I am given a pdf image of a mathematical symbolic formula can I translate it into Mathematica syntax using ML?

I tried using Classify, but I can only translate parts of the equation and I don't know how do to group the pieces together. Moreover, I found the training set can not have an equal sign. (I pasted an image because of the images in the training set).

image of the training set

produce this output:

enter image description here


  • $\begingroup$ Have you tried TextRecognize? $\endgroup$
    – Carl Lange
    May 5, 2019 at 18:43
  • $\begingroup$ @Carl Lange Yes, it does not work, I think it's because a math formula is not simply text and so it had a different training. $\endgroup$
    – Gluoncito
    May 5, 2019 at 18:55
  • $\begingroup$ It's quite likely that Classify won't work for this, unfortunately. If you have a lot of training data (I suppose it's trivial to generate) you could try to train a neural network to do it. Perhaps Classify will work if you use the right FeatureExtraction - but again, only if you have enough training data. $\endgroup$
    – Carl Lange
    May 5, 2019 at 19:00
  • $\begingroup$ I wish I could do it with Mathematica. If the appropriate training model exists in other languages, someone might package it. OCR of math expression $\endgroup$
    – Xminer
    May 6, 2019 at 12:36
  • $\begingroup$ @Xminer, thankx for the link. Foxit has an OCR that partialy recognizes the math formulas, but is far from complete. $\endgroup$
    – Gluoncito
    May 6, 2019 at 13:45


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.