SafeRecall is a CLI tool to OCR and search for text in your photos.
- Built-in OCR: Uses the OS-native OCR engine where available (Apple Vision on macOS, Windows OCR on Windows) and falls back to an embedded ocrs engine everywhere else — no need to install external OCR engines like Tesseract.
- Fast Fuzzy Search: fzf-style fuzzy search through OCR results cached in a local SQLite database.
- Static Binary: Compiles to a single binary with everything included.
recall "text to find" [directory] # index new/changed images, then search
recall -g "text to find" # search everything ever indexed
recall --cached "text to find" # search the cache without re-scanning
recall --wipe # clear all cached OCR resultsDirectories are walked recursively (hidden files and .gitignored paths are skipped), OCR runs in parallel (-n to limit threads), and results are cached — a file is only re-OCRed when its modification time changes.
This tool uses pre-trained OCR models from the ocrs project by Robert Knight:
- Repository: Hugging Face
- Author: Robert Knight
- License: Creative Commons Attribution-ShareAlike 4.0 International (CC BY-SA 4.0)
The models are used unchanged. No endorsement by the original author is implied.
See THIRD_PARTY_NOTICES.md for full attribution of the libraries used.