Closed tshrinivasan closed 4 years ago
Here is the document on how to train for Tamil https://github.com/harish2704/pottan-ocr/wiki/How-to-train-pottan-for-another-language
I have installed it locally. The local version for malayalam is working fine.
But the training instructions for Tamil is not clear.
Will check with the team.
note from harish
it is now possible to run pottan-ocr training using Keras . added Google cloab notebook in the repository so that, any one can instantaneously start a training on Google colab Please check https://colab.research.google.com/github/harish2704/pottan-ocr/blob/keras-training/misc/pottan_ocr.ipynb
The model used in keras version of the ocr is very light weight . ( I removed one conv layer and decreased imageHeight from 32 to 16 ). But any one can change that before starting their own training session
Hi may I know which data is used for Tamil OCR purpose ..
As tesseract is matured, the potten ocr is development is stopped.
Hence, closing this.
https://github.com/harish2704/pottan-ocr
Online demo - https://harish2704.github.io/pottan-demo/
This is malayalam ocr based on machine learning backed by pytorch. Explore this and make it to work with Tamil