MyOCR

Some scripts that I copy from https://github.com/astutejoe/tesseract-tutorial and modify for shan language training (Myanmar sub language) .

Require / Sources

• https://github.com/tesseract-ocr/tesseract (Used validate_myanmar.cpp modified version)
• https://github.com/sengkyaut/tesseract (my repo modified version)
• https://github.com/tesseract-ocr/tessdata_best (For mya.lstm ref)
• https://github.com/tesseract-ocr/langdata_lstm (For charsets)
• https://github.com/kaungyeehein/mm-ocr (tesseract dev script and tools installation)

Installation (for training)

Clone langdata

git clone https://github.com/tesseract-ocr/langdata_lstm.git

Using install script

git clone https://github.com/sengkyaut/MyOCR.git
chmod +x MyOCR/install/install.sh
./MyOCR/install/install.sh

Using Docker

How to install Docker

curl -fsSL https://get.docker.com -o get-docker.sh
sh get-docker.sh

Run Docker (recommend)

git clone https://github.com/tesseract-ocr/langdata_lstm.git
cd langdata_lstm
sudo docker pull sengkyaut/t4cmp:latest
sudo docker run --rm -d -p 2222:22 -v ${PWD}:/langdata_lstm --name skt4cmp sengkyaut/t4cmp

or

git clone https://github.com/sengkyaut/MyOCR.git
cd MyOCR
docker-compose up

login to docker container

ssh root@localhost -p 2222 password is toor

• tesserect source repo and MyOCR repo located in /root/workspace
• langdata_lstm repo also mounted in /langdata_lstm

To do

• Prepare training data
• Retrain or Finetune_plusminusmyanmarmodel with full shan charsets
• Check percent and Finetune other shan fonts