Precense detector and people counter lamp help

Help
#1

Hello again! I want to build a project for my design school project where there is a light fixture that only works when an embedded device detects two or more people in the room. Im thinking a night vision camera as It will be dark in the room if you are only one person present then running perhaps FOMO for inference?

Any tips on how I can achieve this the easiest/cheapest?

Also, as I am new to programming. Would this be easy to make in say python? I seem to recall there are examples files for python on the Edge impulse website, do these work also with fomo? (not that I really need fomo im guessing, as the complexity is not that high)

The thought is that: when one person is detected, nothing happens. When two people are detected, a GPIO pin goes high and controls a relay to turn on the light fixture. As a bare minimum proof of concept. Version two I might try to dim up the light more the more people are present.

Thanks,
J

#2

Hi @Instant_exit ,

You can achieve this using any of the Edge Impulse supported vision boards.

For example if you have a Raspberry Pi / Jetson Nano (4GB/2GB) you can use either FOMO or the standard SSD Mobilenet for training the object detection model, then do the counting and lamp switching using any of the available high level language SDKs:
Python
C++
Go
Node.js

Have a look at this tutorial on how to build your object detection model for the linux clients:https://docs.edgeimpulse.com/docs/tutorials/object-detection

You can also achieve this using FOMO if you have any of the supported MCU targets. Have a look at this tutorial: Counting objects using FOMO - Edge Impulse Documentation

Let me know if this helps.

Regards,
Clinton

1 Like