Menu

Home

fandaoerji
Attachments
fig1.png (14571 bytes)
fig2.png (12816 bytes)
mnglUnicode01.jpg (212915 bytes)

Usage Instructions

1.Download database from https://sourceforge.net/projects/mongolian-hw/files/

  1. File Introduction
    Trainset.mat:Matlab(ver:7.10.0.499 (R2010a)) format train set file.
    Trainset_label.txt : Train set labels in ASCII text format(Unicode)

    Testset_I.mat:Matlab(ver:7.10.0.499 (R2010a)) format test set file I.
    Testset_I_label.txt : Test set I labels in ASCII text format(Unicode)

    Testset_II.mat:Matlab(ver:7.10.0.499 (R2010a)) format test set file II.
    Testset_II_label.txt : Test se II labels in ASCII text format(Unicode)

  2. Operate in Matlab

   load('Trainset.mat');
   imshow(train_set{1});

you will see:

train_set is cell with size 100000x1, each cell is a handwritten binaray image with size Hx48,where H is handwritten image height,48 is fix width.

  1. You can open Trainset_label.txt by any text editor, but you need mongolian font file in https://sourceforge.net/projects/mongolian-hw/files/font/.
    Trainset.mat and Trainset_label.txt corresponds by natural order.
    For example the label of train_set{100} is line 100 in Trainset_label.txt .

4.Mongolian Unicode Table

Reference

Fan Daoerji, Gao Guanglai, "DNN-HMM for Large Vocabulary Mongolian Offline Handwriting Recognition", 2016 15th International Conference on Frontiers in Handwriting Recognition, vol. 00, no. , pp. 72-77, 2016, doi:10.1109/ICFHR.2016.0026

For any question contact me at fandaoerji@126.com


Want the latest updates on software, tech news, and AI?
Get latest updates about software, tech news, and AI from SourceForge directly in your inbox once a month.