uxzView, a free GPL-3.0 image viewer, runs offline Tesseract OCR and ZXing-C++ barcode reading on a right-clicked selection

Right-click a patch of a screenshot in uxzView and the viewer will read the text in it, or decode the QR code in it, without sending the picture anywhere. The recognition runs on your own machine, which matters when the screenshot is an error dialog with a license key in it.

The project’s README describes the app as an image viewer for Windows 10 and 11, Linux and macOS, built with C++20 and Qt 6, and says it is “Currently in development.” The Windows package carries QR and barcode reading through ZXing-C++ and text recognition through Tesseract, with English, German, French, Japanese and Korean language data. The interface itself comes in the same five languages.

The mechanism is deliberately plain. “Tools live in tools beside the application and run only when invoked,” the README says, so nothing scans your images in the background. Recognition works on the whole image or on a selection you right-click, and the tools, the README says, “leave image pixels unchanged,” so reading a barcode doesn’t rewrite your file. If you don’t want a tool, the Options menu has a Manage extensions page under Function extensions where individual tools and OCR languages can be removed or moved elsewhere.

Reading text is the headline, but the rest of the feature list is a full viewer. The README lists “Multi-page PDF reading with page navigation and image export,” folder browsing, fullscreen viewing, batch processing, image editing, ratings, metadata search and duplicate detection, plus RAW, SVG and ZIP/CBZ support (the last through a plugin). It also lists “Built-in jpegli JPEG export,” and says the build downloads pinned jpegli and Highway sources and checks their SHA-256 hashes before it uses them. That makes a pleasant change from “trust me, it’s a zip.”

There is also a portable mode. Turn it on under Options, then Settings, then Save, restart, and the README says to move portable.mode, IxView-data and tools together with the app, which makes the viewer a folder you can carry on a USB stick. (Open source, portable and a folder browser: the “IrfanView, but…” pitch writes itself.)

The catch is how finished it is. The README says prebuilt packages target Windows x64 only, and for Linux and macOS you build from source with CMake 3.24 or newer and Qt 6.8 or newer, including Qt PDF. It then says “successful native Linux/macOS runs are still outstanding.” The Releases section was empty when I read the page, so what you’d run today is a build you make yourself. The repo also keeps a stabilization document dated November 2, 2026, which fits a project still being hardened.

Viewers that read pictures are a natural place for OCR to end up, because a screenshot is where most text gets trapped these days. The same folder-of-tools design could take new language packs or other decoders without the app growing, which is presumably why the README lets you remove tools one by one. Whether the Linux and macOS builds ever get their first confirmed native run is the thing the README itself leaves open.

The tools “leave image pixels unchanged.”