[FEAT] Expand for CAR/BUS to use deepstack-ai with logo-detection (as detector)

[ozett]

Is your feature request related to a problem? Please describe. i want to use another detector in double-take. it has a similar api to the face-detectors, but could detect logos

Describe the solution you'd like i would like to simply configure another detector for the pre-detected categories CAR/BUS (instead of PERSON for faces)

Additional context the api looks familiar to what you have. may not the biggest efford to adopt this custom deepstack-ai for logo-detection . let me know what you think.

[surge919]

+1

this is similar to a question I posted on the home assistant forum. I was asking if I could train an image of a vehicle (ie: the mail carrier), but this would take it to the next level and catch all delivery vehicles. (at least those with a logo)

[ozett]

i turn out that the logo-detection is baseed on yolov5. so it could also be implemented directly or with doods2

https://github.com/snowzach/doods2/issues/28

[hillbicks]

While this is still open, I'm already doing this via node red. There is a deepstack addon for node red, that you can use with the openlogo set. It works only in theory though, because the snapshots are too small to be recognized by deepstack/openlogo.

But I have the same problem with face recognition, so YMMV.