Custom component: image classification via Tagbox

[robmarkcole (Robin)]

This custom component implements image classification via Tagbox. You can create binary sensors for tags in images and use these in automations. For example, you might have an automation which is triggered when your dog enters the living room.
Cheers

robmarkcole/HASS-Machinebox-Tagbox

[robmarkcole (Robin)]

I’ve just released version v0.2 which makes config consistent with Facebox. Note that in the latest release of Tagbox you can now limit classification to custom tags, which should improve performance when configured. Any comments/feedback are welcome.
Cheers

[CountParadox (Lewys Martin)]

This is pretty awesome dude, I like it,

https://blog.lewys.eu/?p=742 set it up just now,

Is it possible to like, specify an area ?

How do I know its going to detect what I want it to?

Does it look for NEW things ?

[robmarkcole (Robin)]

HI @CountParadox for an area (region of interest, ROI), you could open a feature request/PR on https://www.home-assistant.io/components/camera.proxy/

You need to experiment to confirm it will detect what you are interested in.

[ffyter]

Rob, is it possible to run facebox and tagbox at the same time? I tried installing it on docker and I got the port was in use. Obviously because facebox was using 8080. I changed the port to 8181:8181 and I get no errors on install but I am unable to access the instance.

[robmarkcole (Robin)]

HI, just configure tagbox to use 8181 then. Cheers

[ffyter]

I did but the instance is not available when I go to it. It says the site can’t be reached. Oddly when it was installed the text that says it’s up and running says go to port 8080 even though I specified -p 8181:8181.

[robmarkcole (Robin)]

By ‘installed’ are you referring to when you spun up the docker instance? For docker advice the machinebox forums are also helpful

[masterkenobi (Ben)]

Don’t change the container port. Just change the local port. In your HA configuration, use the local port value.

I am running multiple Facebox, Classisficationbox and tagbox in my Synology Docker in such way.

[ffyter]

Thanks! Worked like a charm.

[robmarkcole (Robin)]

@popboxgun I just unarchived and update this integration, so its good to go now

[popboxgun (ha_popbox)]

Much appreciated!

[ahue]

Great thing! I experimented with tagbox today but quite soon hit the teach limit of 100 examples per tag. Do you know an alternative service that runs in docker and supports transfer learning?

[robmarkcole (Robin)]

Checkout Deepstack, which supports training custom obj det models.

Face and person detection with Deepstack - local and free!

Hi all I just published a custom component for face and object (e.g. person) detection using Deepstack, which runs locally in Docker and costs you nothing. You will need a computer with a decent amount of RAM available, as Deepstack is using deep-learning models. Therefore don’t expect to run this on a pi, but a spare laptop should do. Any feedback on usage here, and bug reports on Github. Thanks! [image]