Awesome
<p align="center"> <a href="" rel="noopener"> <img width=100px height=100px src="doc/logo_doc.png" alt="Project logo"> </a> </p> <h2 align="center">Poricom</h2> <p align="center"> Optical character recognition in manga images. Manga OCR desktop application</p>Contents
About <a name = "about"></a>
Poricom is a desktop program for optical character recognition in manga images. Although it is a manga OCR application, it can recognize text on other type of images as well. The project is a GUI implementation of the Manga OCR library (supports Japanese only) and the Tesseract-API python wrapper tesserocr (supports other languages). See demo below to see how it works.
Detect text on locally stored manga images:
Perform OCR on the current screen by pressing Alt+Q:
Alternatives <a name="alternatives"></a>
- Cloe - The app is based on Poricom's global snipping functionality. If you downloaded Poricom to use only the global shortcut, it might be better if you use Cloe instead.
- mokuro - Converts manga images to web pages with selectable text. This saves you time and manual effort since textboxes are automatically detected with an almost 100% accuracy.
User Guide <a name="user_guide"></a>
Follow the installation instructions here. Load a directory with manga images and select text boxes with Japanese text. If you are not getting good results using the default settings, use the MangaOcr model to improve text detection.
Features
Listed below are some of the features of Poricom. Smaller features that are not covered in this section are mentioned in the changelog. Click the arrow to see how each implemented feature works.
<details> <summary>Open a directory with manga images or a supported manga file (cbz, cbr, pdf) and start scanning text bubbles.</summary> </details> <details> <summary> Capture images outside the application using the shortcut `Alt+Q`.</summary> </details> <a name="load_model"> <details> <summary>Load MangaOcr model to improve Japanese text recognition.</a> </summary> </details> </a> <details> <summary>Change language and/or orientation (limited to the Tesseract API).</summary> </details> <details> <summary>Detect text on non-manga images.</summary> </details>Installation <a name = "installation"></a>
Download the latest zip file here. Decompress the file in the desired directory. Make sure that the app folder is in the same folder as the shortcut Poricom.
System Requirements
Recommended:
- Hard drive: at least 800 MB HD space
- RAM: at least 2 GB
Approximately 250 MB of free space and 200 MB of memory is needed to run the application using the Tesseract API. If using the Manga OCR model, an additional 450 MB of free space and 800 MB of memory is required.
Development Setup
- Clone this repo and install conda.
- Install dependencies by running
conda env create -f environment/base.yaml. - Activate the environment with
conda activate poricom-py39and run the app usingpython main.py. - If you want to build the app locally, install build dependencies by running
conda env update -f environment/build.yaml. Then runpyinstaller main.specin thebuilddirectory.
Acknowledgements <a name = "acknowledgements"></a>
This project will not be possible without the MangaOcr model by Maciej Budyś and the Tesseract python wrapper by sirfz and the tesserocr contributors.
The software is licensed under GPLv3 (see LICENSE) and uses third party libraries that are distributed under their own terms (see LICENSE-3RD-PARTY).
The icons used in this project are from Icons8.