I was wondering now if it’s possible to change the FPS of video streams that are displayed at 30 fps, and I’d like to test with 60, and possibly change the display resolution.
The FPS and resolution of source are controlled from outside. However, please note that at certain point you may be bottlenecked by CPU (video decode, resize, display).
Also, is it possible to use 2 AI models per camera? If I want, for example, a model to do multi-human pose estimation by posing joints and a person re-identification model (in case of blind spot), all of these per camera in real time. Do you have an example of code for this, or something similar?
Yes, it is possible to use 2 AI models per camera. These can be in series or in parallel. You can see some guides we published on this topic: A Comprehensive Guide to Building a Face Recognition System, A Comprehensive Guide to Building a License Plate Recognition (LPR) Systems