Home automation project help - People tracking for lighting control

Currently have my entire home automated with tp-link lb100 and lb130 wifi bulbs, tp-link smart outlets, logitech harmony hubs for IR+BT control, a handful of google home minis, nest thermostats, and chromecasts, both audio-only and normal. Sounds first worldly but I’m getting tired of having to ask google to do literally everything.

So, I also have a handful of Wyze cam v2’s, which have custom firmware available that allows for RTSP. I also happen to have a beefy docker host that can run everything needed for my idea. Which is this:

Have lights follow people around the home.

Here’s my criteria:

  1. Ability to identify people vs animals vs the roomba, etc. (facial recognition a bonus)
  2. Send out alerts to either Home Assistant or IFTTT for lighting scene control (like, if a person walks in the kitchen at night, set these bulbs to 3%)
  3. All local processing (no Google Vision/AWS Rekognition)
  4. Dockerize-able, preferably
  5. Open Source, preferrably
  6. Fast. I don’t want any more than a 5s delay in image processing.
  7. Able to be turned off via IFTTT (e.g. “Hey, Google, activate privacy mode”, “Okay, I’ve turned off all the cameras”. Don’t need it to record them night time activities.)

I have found projects like iSpy, which is open source, and can do facial recognition plus people tracking (with a $25 iPlugs license) but windows only. Don’t want to spin up a vm or pay for a license but if I have to w/e.

Also found Facebox, which can integrate with Home Assistant, and can be used for, say, announcing if people are at the door and who they are, which is cool, may use that to welcome people as they walk in, but doesn’t seem to do the activity tracking I want.

Any suggestions?

1 Like