I would look into frigate - it uses openCV and is also written in Python. Very popular with Home Assistant users.
Python
Welcome to the Python community on the programming.dev Lemmy instance!
π Events
Past
November 2023
- PyCon Ireland 2023, 11-12th
- PyData Tel Aviv 2023 14th
October 2023
- PyConES Canarias 2023, 6-8th
- DjangoCon US 2023, 16-20th (!django π¬)
July 2023
- PyDelhi Meetup, 2nd
- PyCon Israel, 4-5th
- DFW Pythoneers, 6th
- Django Girls Abraka, 6-7th
- SciPy 2023 10-16th, Austin
- IndyPy, 11th
- Leipzig Python User Group, 11th
- Austin Python, 12th
- EuroPython 2023, 17-23rd
- Austin Python: Evening of Coding, 18th
- PyHEP.dev 2023 - "Python in HEP" Developer's Workshop, 25th
August 2023
- PyLadies Dublin, 15th
- EuroSciPy 2023, 14-18th
September 2023
- PyData Amsterdam, 14-16th
- PyCon UK, 22nd - 25th
π Python project:
- Python
- Documentation
- News & Blog
- Python Planet blog aggregator
π Python Community:
- #python IRC for general questions
- #python-dev IRC for CPython developers
- PySlackers Slack channel
- Python Discord server
- Python Weekly newsletters
- Mailing lists
- Forum
β¨ Python Ecosystem:
π Fediverse
Communities
- #python on Mastodon
- c/django on programming.dev
- c/pythorhead on lemmy.dbzer0.com
Projects
- PythΓΆrhead: a Python library for interacting with Lemmy
- Plemmy: a Python package for accessing the Lemmy API
- pylemmy pylemmy enables simple access to Lemmy's API with Python
- mastodon.py, a Python wrapper for the Mastodon API
Feeds
Also allows you to use hardware acceleration for inference. Quite a comprehensive set of tools actually, also the new revamped UI is on the horizon with version 0.14
I'm guessing you might be trying to edit the software to do something else as well, in which case it might be possible to embed Python somehow or get it to call another program.
It seems like you want to do some kind of facial recognition. That feels a bit out of my league if I'm honest, but I can tell you the sort of thing I would probably try to do if I had to do that.
You'd want some footage from when someone's face was close to the camera, so you could try getting snapshots of when there was a lot of motion in the video, then save those to some folder as pictures.
Once you have those pictures, it sounds a lot easier to try to run some complicated facial recognition algorithm on them than just "using the right camera software".
Presumably the CCTV is controlled by some server, so I would get the server to then handle the facial recognition separately, either on a schedule or soon after the motion close to the camera. An underlying facial recognition software could hopefully be called as part of a shell script on the server.
If that's the problem you're trying to solve, I think I broke it down into a few subproblems, which might help. I might be a bit wrong though because I have no experience doing anything with CCTVs.