Take a picture of a page and get clean text—recognize on your device, preserving the original layout.
YueShi turns any page into readable text. Point your camera, select an image from your album, or drag in a PDF, and Android ML Kit recognizes printed text on your phone or tablet—fast, free, and available offline; the original image never leaves your device.
What it can do:
• Recognize on your device—Android ML Kit recognizes printed text on your device, free, no registration required, and the image never leaves your device.
• Preserve the original layout—Text is returned exactly as it appears on the page, with columns, tables, and formulas in their proper places, preserving structure instead of being compressed into a bunch of characters.
• Multiple text types, choose your recognition engine—Printed text, mixed languages, and handwriting are all supported. Local recognition is available, or for the most challenging layouts, you can connect to cloud engines: Azure, Mistral, TextIn, and multimodal models such as Claude, GPT, and Qwen.
• Your Key Remains Yours – Any cloud engine can come with its own key. The credentials and recognized text remain solely in your account, never entering a public pool or being resold. You pay the service provider directly, with no markup.
Buy Once, Own Forever
Free to use, perfectly adequate for daily needs. Want everything? Buy now for a lifetime membership – no subscription, no renewals.
• Trial/Registered Version: 5-10 pages per file, export to PDF and Markdown
• Lifetime Membership: Unlimited pages, up to 10 parallel tasks, unlimited service credentials with polling support, export to DOCX and raw parsed data
Privacy First
Local recognition without leaving the device. The cloud engine uses your own API key to connect directly from the device to the service provider – neither the file nor the key passes through our servers. We do not save your images, documents, or recognized text.
Let the text transcend paper.
Contact Us: support@kaichenera.com
Local OCR recognition ensures columns, tables, and formulas are all correctly positioned.