Oculex (formerly EOES) is an OCR Stream Aggregator. It ingests video streams (e.g. RTSP), extracts text via OCR, and provides unified output. Designed to integrate nicely with Home Assistant for automations, monitoring, and logging.
- Overview
- Features
- Architecture & Tech Stack
- Installation
- Configuration
- Usage
- Integration with Home Assistant
- Limitations & Considerations
- Roadmap
- Contributing
- License
Oculex acts as a centralized node for OCR processing:
- Collects live video streams (e.g. RTSP from IP cameras).
- Extracts text from these streams using OCR.
- Normalizes and standardizes the output so downstream systems (like Home Assistant) can consume it easily.
- Enables automations or alerts based on recognized text.
- Stream Ingestion — Support for RTSP and potentially other stream protocols.
- Frame Sampling and OCR — Periodic or event-triggered frame captures for OCR processing.
- Unified Output Format — Text output normalized; includes metadata like timestamp, camera/source id, confidence, bounding boxes.
- Flexible OCR Engine Options — Possibility to use different OCR engines (e.g. open source like Tesseract, EasyOCR, etc.).
- Home Assistant / Smart Home Integration — Expose outputs as sensors, events, or via API/MQTT so Home Assistant workflows can use them.
- Containerization / Docker Support — Easier deployment & portability with support as a home assistant add-on.
| Component | Role |
|---|---|
| Stream listener / grabber | Connects to RTSP (or other) sources, pulls frames. |
| OCR Processor | Applies OCR on captured frames (configurable engine). |
| Output formatter | Normalizes results (timestamp, source, confidence, bounding box etc.). |
| API / Interface / Broker | Sends results to consumers (Home Assistant, MQTT, Webhooks etc.). |
| Configuration layer | Controls which streams, sampling rate, thresholds, OCR engine etc. |
| Deployment | Docker / container setup, optionally local or cloud deployment. |
As this project is still in early alpha, there are no installation steps yet. However, you can open an issue and I will help you get this set up :)