I’ve been experimenting with presence detection for quite some time now and have an LD2450 connected to an ESP8266 installed in every room. The detection itself works great, but the “zoning” is a rather tedious and challenging task, especially when setting up the zones. Anyone who has played around with this knows what I mean. For example, I have an L-shaped hallway and want to define each corridor as a separate zone—ideally as many small zones as possible so the lights can follow you. 
It’s been a while since I was in school, but some things from math class have stuck with me, and I came up with the following thought experiment. I’d love to hear your opinions on it:
If I can get the coordinates of detected people (target sensor x, y) through HomeAssistant, can’t I define zones within HomeAssistant and check whether a target is inside a polygon and how many are there?
→ Here’s how I imagine it: you measure four points (the corners of your zone).
From these, you can derive four individual 𝑓(𝑥) functions, which define an area and thus a zone/polygon. Then you check which areas or planes the target coordinates fall into. In my view, this would allow you to divide the space into as many shapes as you like.
This way, you could create binary sensors that indicate whether a zone is occupied and how many people are in it.
I’ve also made an initial attempt, but it’s not quite working yet—what do you think?
- name: TestZone
unique_id: test_zone
icon: mdi:calendar-today
state: >
{% set x_target = states('sensor.sens_sz_presence_p0x') | float %}
{% set y_target = states('sensor.sens_sz_presence_p0y') | float %}
{% set polygon = [(0, 0), (0, 500), (500, 500), (500, 0)] %}
{% set count = 0 %}
{% for i in range(polygon | length) %}
{% set x1, y1 = polygon[i] %}
{% set x2, y2 = polygon[(i + 1) % (polygon | length)] %}
{% if (y1 > y_target) != (y2 > y_target) and
(x_target < (x2 - x1) * (y_target - y1) / (y2 - y1) + x1) %}
{% set count = count + 1 %}
{% endif %}
{% endfor %}
{{ count % 2 != 0 }}
##############################################
edit
I found out why it didn’t work – the variable is reinitialized after each loop and is therefore reset every time.
So I wrote it in Python, and voilà – it works so far. I can define a polygon with 4 vertices and use the ray casting algorithm to check whether the target point is inside or outside the polygon. It works quite well and is very fast.
Now I’m extending the script to handle a number n of polygons. With automation, you can easily trigger the script for every change to the targets coordinates, making it respond to updates automatically.
target_x and tsrget_y are just the entities of sensor_presence_p0x oder p0y …
I just need to figure out how to use the response in the most meaningful way. I was considering using a Boolean helper or something similar. any good ideas?
Here’s the Python script:
target_x = float(data.get("target_x", 0))
target_y = float(data.get("target_y", 0))
polygon = [(0, 0), (0, 500), (500, 500), (500, 0)]
def point_in_polygon(x, y, poly):
count = 0
for i in range(len(poly)):
x1, y1 = poly[i]
x2, y2 = poly[(i + 1) % len(poly)]
if (y1 > y) != (y2 > y) and (x < (x2 - x1) * (y - y1) / (y2 - y1) + x1):
count += 1
return count % 2 == 1
#Prüfen, ob Punkt im des Polygon liegt
in_zone = point_in_polygon(target_x, target_y, polygon)
hass.services.call("notify", "persistent_notification", {"message": f"Person ist {in_zone}. {target_x} {target_y}"})
