Architecture Proposal Integrating PostgreSQL for ML and Surveillance Scalability

[Moin2002-tech]

I’ve been thinking about the next steps for the Smart Store surveillance system, and I really think we should go with PostgreSQL for our database. Since you’re already using OpenCV and Boost for the heavy lifting, Postgres would be a huge help because it’s specifically designed to handle the kind of complex data machine learning produces. Instead of just storing raw text, we can use the pgvector extension to store and query the actual mathematical 'fingerprints' of the objects or faces the cameras detect, which makes searching through hours of footage way faster. It also handles multiple streams of data at once without breaking a sweat, so as we scale up the number of cameras in the store, the database won't become a bottleneck. It’s basically the most 'production-ready' way to bridge your C++ processing logic with a reliable data layer that can actually grow with the project.

[gem870]

I have considered that, and that is what I am going to use as the project grows. But now I started with SQLite for now, I will switch once the system starts having heavy computation, and with time I will have to restructure the surveillance system especially the database.
I am trying to make it product ready so that when users start using it, from the user's feedback, I would know what to do next.

Currently I am working on the UI, Smart_Store will be the power engine for the backend.