What framework to use for the camera to detect the number of people?

Question

Python

I want to try something with a neuron.

The task is this: through a camera with raspberry, determine in real time the number of people passing by in the frame.
With maximum settings, so that you can also determine how far / close a person passes from the camera.

In general, if *2 people pass by
, I send one request.
* 3 people, I'm sending another request.
* 4 people, I send the third request.
and so on

And so
* 2 people, one far away, the second nearby, I send one request.
* 3 people, one far away, two nearby, I send a second request.
* 4 people, two far away, two nearby, I send a third request.
etc

So that it would be possible to immediately hang some kind of ID on a person and track him as he goes, so that every second while he is in the frame, send a request. But not a specific person, but just an image.

Give at least directions on which neural networks to dig, any language in general, better than python, js, but without deepening, you can do the rest. And will the raspberry cope with this task and will the xiomi camera fit all this

Answer 1

YOLO
https://habr.com/ru/post/503200/
MEDIAPIPE
https://habr.com/ru/post/596043/
There are others too
https://github.com/abhineet123/Deep-Learning-for-T ...
and Google: open source neural network for people tracking
github:
https://github.com/search?q=people+tracking
https://github.com/search?q=human+tracking&type=