I have used the MNIST data set many times to train models for digit recognition based on object character recognition (OCR).

I am now trying to do the same but with a data set of svg paths.. I am trying to find an MNIST equivalent of a digital path / svg based data set.

Here is a sample:

enter image description here

the svg

<path d="m233.5,119.4375c-1,-1 -3.025818,-1.320366 -5,-1c-3.121445,0.506538 
-8.191559,0.090805 -15,2c-14.665848,4.112541 -23.266006,8.139008 -31,11c-6.291519,
2.327393 -11.679474,6.571106 -14,11c-1.467636,2.801086 -2,7 -2,10c0,4 
-0.610916,8.03746 0,13c0.503769,4.092209 2.877655,8.06601 4,10c1.809723,3.118484
 4.718994,6.310211 8,9c5.576645,4.571762 11.887314,5.376694 18,7c9.8564,2.617508 
19,2 34,2c11,0 19,0 24,0c5,0 9.222717,-1.723984 13,-5c2.136749,-1.853195 4.346191,
-4.70546 6,-7c1.307465,-1.813995 1.693512,-6.048325 1,-11c-1.009766,-7.209747 
-3.793945,-12.087433 -6,-17c-1.832031,-4.079666 -2.714111,-7.21167 -5,
-10c-1.793182,-2.187347 -5.714111,-5.21167 -8,-8c-2.689789,-3.281006 -4,-5 -6,
-7c-1,-1 -4,-3 -5,-4c-1,-1 -2.042908,-1.710213 -3,-2c-3.450851,-1.04483 
-3.852737,-2.173096 -5,-3c-1.813995,-1.307449 -3,-2 -6,-2c-2,0 -3.878555,
-1.493462 -7,-2c-0.987091,-0.160179 -2,0 -4,0l-1,0" id="svg_1" stroke-width="1.5"
 stroke="#000" fill="none"/>

I intend to train the model to understand the d= param. MyScript uses this approach rather than OCR, https://developer.myscript.com/. They call their approach iink but is a brand rather than a word I can use to search for material in this space. MyScript works with 2 dimensional data, so not only just the paths but also the relative positions of objects. I am interested in applying this approach to mathematics but for now anything will do.


I managed to fine some stuff

I then found out about 'CROHME' (Competition on Recognition of Online Handwritten Mathematical Expressions) this also uses InkML.


Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.