Huskylens Homeassistant

Share your Projects!
1

I wanted to use HuskyLens with Home Assistant—specifically for face recognition—and since there is no official integration,opted for what is likely the easiest solution: connecting via MQTT. The HuskyLens sends data to an ESP32 via I2C, and the ESP32 then forwards it via MQTT. I used the HuskyLens Arduino library as a starting point.

husky
husky1479×540 39.3 KB

2

Tried this out - just to play with it. Works great. Thanks… still not sure what to do with it… but it works reliably!
Thanks again.

3

You can use facial recognition for things like unlocking doors or automations targeted at specific individuals; Husky recognizes the face, sends the info to Home Assistant, and Home Assistant knows exactly who it is.
advantage, recognition takes place directly on the husky, no other application, no integration is needed, no demands on your hardware

4

If you want to see what the Husky is capturing, you will need to add a camera that sends photos to the HA system based on data from the Husky; it is not possible to get images directly from the Husky itself

5

Here are a few other things you can do with HuskyLens:

  • It recognizes colors, tags, QR codes, and barcodes, enabling various automation tasks.
  • It recognizes objects, opening up further automation possibilities; Husky identifies an object and triggers an action. For example, you can teach Husky what a coffee mug looks like, and when it spots the mug on the table in the morning, it could trigger a morning briefing to accompany your coffee; or teach it to recognize a cooking pot so that when it sees one on the stove, it turns on the range hood, and so on.
    Most of these automations could technically be achieved without Husky using cameras you might already have at home. However, the advantage of Husky is that all recognition processing happens directly on the device itself, with data sent out via I2C or UART. With standard cameras, you would need a model to handle the recognition—either running locally (which is hardware-intensive) or via a cloud-based model and integration (like OpenAI or Google), which involves service fees.