AaronGIG/pdf2zh-desktop

What it solves

It simplifies the process of translating academic PDF documents while preserving the original layout, formulas, and charts. It removes the technical barriers of the original command-line tool by providing a standalone desktop application that requires no Python installation or environment configuration.

How it works

Built upon the PDFMathTranslate engine, the application allows users to drag and drop PDFs into a graphical interface, select a translation service, and generate a translated version. It supports various AI models (via API or local deployment) and includes specialized pipelines for handling complex elements like tables and scanned images using OCR.

Who it’s for

Researchers, students, and academics who need to translate technical papers quickly without dealing with coding environments or losing the professional formatting of their documents.

Highlights

• Zero-Configuration: Standalone executable for Windows and macOS with all dependencies bundled.
• AI Integration: Supports 20+ translation services including DeepSeek, GPT-4o, Gemini, and local options via Ollama.
• Zotero Integration: Deep linkage allowing users to drag items from Zotero or right-click PDFs within Zotero to translate and automatically save the result as an attachment.
• Advanced PDF Handling: Supports 1000+ page documents via chunked translation, handles scanned PDFs with RapidOCR, and features a dedicated translation pipeline for tables.
• Flexible Output: Offers three layout options: mono (translation only), dual (bilingual), and side-by-side (original on left, translation on right).