naptha/tesseract.js

Tesseract.js is a JavaScript library that performs OCR, Optical Character Recognition, meaning it reads text out of images, directly in the browser or on a Node.js server, without sending data to any external service. The problem it solves is that extracting text from a photo, screenshot, or scanned document traditionally required a server-side process or a native application. Tesseract.js brings that capability entirely to JavaScript, so a web app can let users upload an image and get back the text locally, with no backend required. How it works: Tesseract.js is a JavaScript wrapper around the original Tesseract OCR engine, an open-source recognition system originally developed at HP and later maintained by Google. The wrapper compiles Tesseract to WebAssembly (a format that lets native code run inside a browser at near-native speed) and exposes it through a simple asynchronous API. You create a "worker" (a background processing thread), pass it an image URL or file, and it returns the detected text. Workers can be reused across multiple images to avoid repeated initialization overhead. The library supports over 100 languages, and language data files are downloaded on first use. You would use Tesseract.js when building a browser-based tool that needs to read text from photos, for example, a Chrome extension that extracts text from screenshots, a document scanning webapp, a receipt parser, or a form that auto-fills from a photographed card. On the server side, Node.js 16+ is supported for batch-processing workflows. The tech stack is JavaScript running in browsers (via script tag, ESM, or bundled with webpack) and Node.js. The core OCR engine is compiled to WebAssembly and loaded at runtime. The library is on npm as tesseract.js.