TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Show HN: Improving accessibility of websites by providing image OCR in-browser

2 点作者 LinguaBrowse超过 6 年前

1 comment

LinguaBrowse超过 6 年前
This is an upcoming feature for my personal project, LinguaBrowse: an app for browsing the foreign-language web. It principally applies transcriptions (such as pīnyīn to Chinese, and furigana to Japanese) to text, but also provides dictionary lookup for words with a single click. This demo is for the [macOS version](<a href="https:&#x2F;&#x2F;itunes.apple.com&#x2F;gb&#x2F;app&#x2F;linguabrowse&#x2F;id1422884180?mt=12" rel="nofollow">https:&#x2F;&#x2F;itunes.apple.com&#x2F;gb&#x2F;app&#x2F;linguabrowse&#x2F;id1422884180?mt...</a>), although an [iOS version](<a href="https:&#x2F;&#x2F;itunes.apple.com&#x2F;gb&#x2F;app&#x2F;linguabrowse&#x2F;id1281350165?mt=8" rel="nofollow">https:&#x2F;&#x2F;itunes.apple.com&#x2F;gb&#x2F;app&#x2F;linguabrowse&#x2F;id1281350165?mt...</a>) also exists.<p>The macOS version is built in React Native, using the react-native-macos community project. I write the code in Swift (for native) and TypeScript (for the React Native JSContext). Ideally I&#x27;d have liked to have delivered a click-to-OCR feature, but saving a copy of the file (hence the drag-and-drop) was the only way I could find to overcome both:<p>* CORS issues, with loading the image data into Tesseract; and * the App Transport Security policy, which prevents mixed content in-browser and HTTP downloads by NSURLSession.<p>As a bonus, though, this methodology supports OCR of files dragged in from the file system.<p>Happy to talk about any part of it. Always striving to make foreign-language websites more readable.