AaronGIG/pdf2zh-desktop

What it solves

pdf2zh-desktop is a desktop application designed to simplify the translation of academic PDF documents. It removes the technical barriers of the original command-line tool by providing a graphical user interface (GUI), eliminating the need for users to install Python or configure complex environments. It specifically addresses the-common struggle of maintaining original layouts, formulas, and tables during translation, while integrating deeply with research management tools like Zotero.

How it works

Based on the PDFMathTranslate project, the application processes PDFs by analyzing their layout and using AI translation services to replace text while preserving the original formatting. It supports a wide range of translation providers (from free services like Google and Bing to LLMs like DeepSeek, GPT-4o, and local models via Ollama). For scanned PDFs, it employs a built-in OCR pipeline (RapidOCR) to recognize text and render translations on a clean white background to avoid overlapping with original images.

Who it’s for

Researchers, students, and academics who need to translate technical papers and documents while keeping the original structure intact and who prefer a "plug-and-play" experience over command-line interfaces.

Highlights

• Zero Configuration: Fully bundled Python runtime; works immediately after extraction on Windows and macOS.
• Zotero Integration: Allows dragging items from Zotero for translation and automatically writes the translated PDF back as an attachment.
• Advanced PDF Handling: Supports documents over 1,000 pages via chunked translation and intelligent memory management to prevent crashes.
• Scanned PDF Support: Built-in OCR for image-based PDFs with a dedicated pipeline to handle tables and figures without distorting them.
• Flexible Output: Offers three formats: pure translation, bilingual (interleaved), and side-by-side (original on left, translation on right).
• Extensive AI Support: Pre-configured presets for 20+ models including DeepSeek, GPT-4o, Gemini, and local Ollama deployments.
• Academic Tools: Includes built-in terminology libraries for Computer Science, Medicine, Finance, Law, and Electrical Engineering.