The MIT team has recently open-sourced PDF Craft, a high-quality PDF conversion tool designed specifically to solve challenges in processing technical documents and academic papers. Based on DeepSeek-OCR technology, this tool can accurately recognize scanned PDFs, perfectly restore LaTeX mathematical formulas, and intelligently preserve complex layouts such as double columns and mixed text-image formatting, effectively addressing the pain points of traditional conversion tools. PDF Craft supports Markdown and EPUB format output, and automatically generates tables of contents and annotations. Users can choose to run it locally for free (requiring an RTX 3060 or higher graphics card) or use a pay-as-you-go cloud service. As a completely open-source project (MIT license), users can review the code, deploy it themselves, or contribute to the project. The project provides an online demo and API documentation, is undergoing rapid iteration, and plans to add table support functionality this week.
Original Link:V2EX Share & Discover

IT资源栈
评论前必须登录!
立即登录 注册