Screen capture language content for quick dictionary look ups.
This is a prototype for LangOCR, a cross-platform OCR application for quick dictionary look-ups during active immersion of non-selectable text.
This program is designed in line with the methodologies of the MIA. For more information, please check out the MIA Website.
At this moment, this application is a prototype. The architecture is still being developed so the code is in a high level of flux.
This application is intended to supplement active immersion while reading in the target language.
Yomi-chan and applications like it are fantastic when the text is selectable, but it's difficult to do the same when the text of the medium cannot be selected.
Kanji-tomo satisfied me for a long time, but doesn't have a modular dictionary interface. I also had to copy my sentence cards into anki by hand.
This is where LangOCR comes in.
The goals of LangOCR are to do the following.
- Minimize time the user spends looking up entries in a dictionary
- Minimize time the user spends making N+1 sentence cards for Anki.
- Utilize a flexible dictionary interface for cross-language translations (e.g. EN to JP and JP to JP Translations).
- Log frequency of the words the user comes across.
- Leverage the machine learning engine at the core of this application for various future applications.
- Create a low-load intensity architecture that can capture and decode text over time. This goal is targeted to capture from games that use text as it's main form of communication.
- This application is designed for Windows, Mac, and Linux.
- Implement a flexible 'rule set' so the user can control how LangOCR operates.
Until I have time to test on multiple platforms, I'm going to have to assume that developers interested on contributing can figure out the nitty-gritty of getting the packages they need. The following packages are installed on my system.
- g++ (Arch Linux 9.3.0-1) 9.3.0
- GNU Make 4.3
- QMake version 3.1
- Qt version 5.14.2
- tesseract 4.1.1
- leptonica-1.79.0
Please follow Google C++ Coding Style. All PRs should adhere to Google coding style.
There is an emphasis on readability and modularity. As the project develops, it is realistic that major aspects of the code base will be interchanged.
You can contact me through my twitter, by github, or by email. Email will get the quickest responses.
Skylar Scott – Twitter:@GeekTeaa Email: skyenet.inbox@gmail.com
PR and bug reports should be file on GitHub.
This code is distributed under the GNU license. See ''LICENSE'' for more information.
Inorder for this code to work, you must have a compositor running! You will see a big black screen blocking your view. Windows, Mac, and most Linux users will not see this issue.
If you want to contribute, please and thank you! While the project is in it's early stages, I ask that you contact me before tackling major changes so the changes are in line with the architecture goals.
My speciality is in embedded C and networking programming concepts. Programmers of all backgrounds are welcome, but this project is missing programmers with a strong foundation in QT and Machine Learning.