Repository navigation
Troubleshooting and FAQ
This is almost always the one-time setup on first launch failing to download. Check your internet connection and click Try again. The error screen shows where the logs are:
-
setup.log: setting up DocBox's Python runtime. -
backend.log: DocBox's background service.
If it keeps failing, open an issue and attach both logs. They don't contain the text of your documents.
Check your connection and press the button again to retry. You can also pause a download and resume it later.
Start the Ollama app, or click Start Ollama under Connections, then Check again.
Press Check again under Connections. On Windows, DocBox also looks in Tesseract's
default folder (C:\Program Files\Tesseract-OCR), since its installer doesn't add itself
to PATH. If you installed it somewhere else, add that folder to PATH and restart
DocBox.
| System | Folder |
|---|---|
| Windows | %LOCALAPPDATA%\io.github.docboxai.docbox |
| macOS | ~/Library/Application Support/io.github.docboxai.docbox |
| Linux | ~/.local/share/io.github.docboxai.docbox |
Models › Engines on disk shows it too. Its history folder holds the text of files
you've read. Saved results go to Documents/DocBox unless you choose another folder in
the settings.
No. OCR runs on your computer. DocBox only goes online to download its runtime, engines and models, and to check GitHub for a new version. The one exception is the NVIDIA NIM engine, which works only while the Cloud engine switch is on and you've added your own API key. There's no account, telemetry or analytics.
DocBox needs an Apple Silicon Mac (M1 or later): PaddleOCR, its main engine, has no build for Intel Macs. EasyOCR also needs macOS 14 or later.
Not yet. Both are on the Roadmap.
Using DocBox
Working on DocBox