Manga Ocr

🚀 Manga OCR

Optical character recognition for Japanese text, primarily focused on Japanese manga.

This project leverages the Vision Encoder Decoder framework to perform optical character recognition on Japanese text, with a special emphasis on Japanese manga. While it can serve as a general - purpose printed Japanese OCR, its core objective is to offer high - quality text recognition that can withstand various manga - specific scenarios, including vertical and horizontal text, text with furigana, text overlaid on images, a wide range of fonts and styles, and low - quality images.

The source code can be found here.

✨ Features

• Versatile Text Orientation: Supports both vertical and horizontal Japanese text.
• Furigana Compatibility: Capable of recognizing text with furigana.
• Image Overlay Resistance: Can accurately recognize text overlaid on images.
• Font Adaptability: Handles a wide variety of fonts and font styles.
• Low - Quality Image Support: Provides reliable recognition even for low - quality images.

📄 License

This project is licensed under the Apache 2.0 license.