Swift OCR microservice using Vision for on-device text extraction
Many macOS developers need OCR without sending data to external services. VisionOCR delivers a native Swift server that leverages Apple’s Vision framework for fast, on‑device text recognition. It exposes a LiteParse‑compatible HTTP API, health checks, and Prometheus‑style metrics, all wrapped in an actor‑based concurrency model with graceful shutdown. The result is a ready‑to‑run, self‑hosted OCR service that avoids third‑party dependencies and runs efficiently on Apple Silicon.
View on GitHub →giautm/vision-ocr