MACHINA is a CCTV viewer that uses LLAVA YOLO 11 OpenCV for object tagging in real time, a work in progress. It connects to a high-resolution RTSP stream, queues frames into memory for processing. YOLO assigns specific IDs to objects based on coordinates, size, and timestamp, iterating to match objects. Another thread continuously iterates through object array, making requests to Ollama server for object tagging. The script includes a 16px tolerance for object matching and creates new objects as a last resort. The project aims to create a complete headless security system using modern vision models. Contributions are welcome, and donations are accepted.
https://github.com/PsyChip/machina