Universal Japanese OCR popup dictionary for Windows, Linux, and macOS, meikipop enables instant, frictionless dictionary lookups of on-screen Japanese text. It targets users reading Japanese content across diverse applications like games, videos, and manga, offering a seamless experience beyond browser extensions by leveraging OCR technology.

How It Works

meikipop employs Optical Character Recognition (OCR) to read Japanese text directly from screen captures, supporting full-screen or user-defined regions for efficiency. Dictionaries are pre-processed for rapid lookups, and the UI is designed for responsiveness. A key advantage is its pluggable OCR backend architecture, allowing users to select between high-accuracy remote services (like Google Lens) or fast, private local OCR options, adapting to hardware and privacy needs.

Quick Start & Requirements

• Install: Easiest via prepackaged binaries from https://github.com/rtr46/meikipop/releases/latest. Alternatively, install via PyPI (pip install --upgrade meikipop) for Python 3.10+ environments, or use an editable install for development.
• Prerequisites: Python 3.10+ is recommended. macOS users must grant Input Monitoring, Screen Recording, and Accessibility permissions. Linux Wayland support is alpha and may require additional OS dependencies (e.g., libxcb variants, pygobject) and troubleshooting. Refer to https://github.com/rtr46/meikipop/issues/43 for Python 3.14 workarounds.

Highlighted Details

• Universal OCR: Captures Japanese text from any screen source, including games, videos, manga, and websites.
• Pluggable OCR: Supports local OCR engines (meikiocr, Chrome Screen AI) and remote services (Google Lens), with OWOCR integration for broader backend choice.
• Customization: Highly configurable hotkeys, themes, colors, layout, and scan regions (full screen or selected area).
• Dictionary Management: Can build its own dictionary or import Yomitan dictionaries.

Licensing & Compatibility

Licensed under the GNU General Public License v3.0 (GPL v3.0). This strong copyleft license requires derivative works to be distributed under the same license, potentially impacting integration with proprietary software or commercial closed-source applications.

Limitations & Caveats

meikipop is not designed as an SRS-mining tool and lacks automatic flashcard creation features. The default local OCR (meikiocr) exhibits poor accuracy for vertical text. Linux Wayland support is in alpha and may require significant user effort and additional dependencies. Potential compatibility issues exist with Python 3.14.