(Last Updated: June 5, 2020)

AI/deep learning implementation: Shifting gear in home automation



  • With all the problems associated with device control though zwave, zigbee or local APIs resolved, outgrowing/dumping vera/ezlo, mobile app and cloud dependencies, I have shifted gear to focus on using AI for home automation.

    The first step is on video stream recognition of various objects, ranging from boxes, cars, animals and faces, and use them as triggers for home automation. The next step would be to implement local voice recognition for commands, down to maybe triggering different actions depending on the voice recognized even for the same command. Eventually maybe even being able to create "scenes" based on predicting what one would want but we are a long way from that.

    Thoughts and ideas anyone?

    The first step for me is to get hardware and it seems the most cost effective way at this point is to get a GPU and pass it through to the VM which is running all home automation system. These article helped me decide:

    Not sure what I will implement from here but I needed to start somewhere. 😉
    My 2060 Super finally shipped. The GPU market, even used, is really a bit insane at the moment.


Log in to reply