A Synthetic Dataset for Clustering Handwritten Math Expression TUAT (Dset_Mix)

Research Tasks

Handwritten Mathematical Expression Clustering

2020-07-30 (v. 1)

Contact author

Vu Tran Minh Khuong

Tokyo University of Agriculture and Technology


+81 070 4445 9674

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported License.


The objective of handwritten mathematical expression clustering problem is to group the handwritten math expression patterns into groups of similar ones. By this way, the user can perform an action on a group of similar patterns. This is useful for the marking problem since the teacher can mark and give feedback efficiently.


We use the purity and the marking cost function presented in the following paper to evaluate the methods.

V. T. M. Khuong, H. Q. Ung, C. T. Nguyen, and M. Nakagawa, "Clustering Offline Handwritten Mathematical Answers for Computer-Assisted Marking," Proc. 1st International Conference on Pattern Recognition and Artificial Intelligence, pp. 121-126, Montreal, Canada, 2018.

Dset_Mix.rardata(8 MB)1The ".inkml" files are stored in the "Data_inkml" folder. The ".png" files are stored in the "Data_img" folder.
ICPRAI_2018_Final_114.pdfarticle(629 KB)2The evaluation protocol is presented in this paper.


No comments on this dataset yet.

Add your comment

In order to comment on a dataset you need to be logged on
Register Now!


In order to rate this dataset you need to be logged on
Register Now!