All the images are uploaded on Docker hub in docker-namsel-ocr public repository.

Running the container

Put all the scan images in a directory in your local computer [PATH] and run the image in the background with the latest image version:

docker run -itd --name namsel -v [PATH]:/home/namsel-ocr/data thubtenrigzin/docker-namsel-ocr:latest bash

For example : Path = ~/data and latest tag version = latest

docker run -itd --name namsel -v ~/data:/home/namsel-ocr/data thubtenrigzin/docker-namsel-ocr:latest bash

Running the scripts within the container:

Preprocessing

Scantaillor will prepare all the images stored in your local directory ~/data.

It is possible to add optionaly the threshold value and the double page layout by adding "l2".

docker exec namsel ./preprocess [threshold value] l2

Recognition

Pecha format

Namsel will start the recognition based on a default configuration. It uses ~/data/out directory (the scantaillored images). The result file "ocr_output.txt" will be move to your working directory ~/data

docker exec namsel ./pecha

Book format

Namsel will start the recognition based on a default configuration. It uses ~/data/out directory (the scantaillored images). The result file "ocr_output.txt" will be move to your working directory ~/data

docker exec namsel ./book

Using your own configuration with parameters

Namsel will start the recognition based on a default configuration. It uses ~/data/out directory (the scantaillored images). The result file "ocr_output.txt" will be move to your working directory ~/data

docker exec namsel ./namsel-ocr [parameter1 parameter2 etc...]

Automatising the recognition with the preprocess included

An all in one button for the book and pecha recognition.

The threshold preprocess value can be optionaly add as a parameter and the double page layout by adding "l2".

For the book recognition:

docker exec namsel ./1book [threshold value] l2

And for the Pecha recognition:

docker exec namsel ./1pecha [threshold value]

Sources of docker-namsel-ocr

Please refer to namsel-ocr repository on Github for the source of the base built Docker image.

All the Docker source will take place on docker-namsel-ocr repository on Github.

Realease notes:

v2.2.0 or latest

• add the preprocess argument "l2" for the double layout book
• check if file or directory exist before the deleting or moving actions
• stability improvements
• The script ./namsel-ocr doesn't delete the directory ./data/out after the ocr completion

v2.1.0

• delete the directory "out" after the recognition completion
• test if the "out" directory exists and uses the non-scantailored scan image if the preprocess has not been launched before the recognition
• use the tag "latest" for the basic image

v2.0.0

• add an "all in one button" for the book and pecha recognition, including the preprocessing stage
• add the threshold parameter as an option for the preprocess
• correcting an issue in book script file letting the book recognition work properly

v1.0.0

First release of the project