Skip to content

Latest commit

 

History

History
27 lines (17 loc) · 979 Bytes

File metadata and controls

27 lines (17 loc) · 979 Bytes

Javanese-Sundanese OCR

This project purpose is for complete "Project Telematics" subject on Computer Engineering major, Institut Teknologi Sepuluh Nopember Surabaya.

But we intend to develop furtermore, Because of on our viewpoint javanese-sundanese script is almost extinct. We're very open on collaborate for this project.

changelogs (aksara jawa)

v.alpha.1.0.0

  • Initial commit, no testing at all system
  • Detection work only on basic character and some of vocal character
  • Model use my own architecture adapted from resnet50

v.alpha.2.0.0 (current version)

  • Detection work on all character of javanese script, But use font "tuladha jejeg"
  • Can detection multiple languange (Bahasa + Javanese)
  • Accuracy for some character still bad, Somehow one character detect multiple times
  • Model use Resnet50 and LTSM

changelogs (aksara sunda)

Kindly to daily check this repo for update

For instalation and setup you can read docs folder