Color recognition, Object recognition, or Text recognition camera input as a trigger

Has anyone used a camera with either object, text, or color recognition as a trigger in any of their projects. I’ve got a unique challenge for a children’s museum that I’d like to find a solution for, without reinventing an existing project if possible. Thanks!

Not sure there’s a direct solution, but an alternative could be QR

1 Like

Hey @cullenb

Send me a PM, I actually did something similar using a webcam and the TUIO protocol to detect objects when in front of a camera.

Update: Just checked it was actually OpenCV but you can do it with TUIO on a webcam.

What do you need to do once it’s detected?

Kind Regards

Louie

1 Like

Thanks, Melvyn. That’s a great idea. I’ve got to set up a scenario where there’s 8 possible acceptable combinations of 6 QR codes together out of 6.2 x 10^23 possible combinations. Fun.

ha…you lost me with those numbers

We’re still trying to iron out the details, but initially we’re thinking of having two short rectangular towers with 3 rotating cubes at the top of each. When 3 cubes match in terms of the color of the faces (on one of the two towers), it triggers content on a flatscreen on the wall behind the 2 towers.

This is where the factorial of 24 came into play with the possible combinations. Each side alone has combinations to the factorial of 12, which is over 479 billion possible combinations where only 4 are recognized as triggers. Ugh.

Hi @cullenb,

Not sure it would fit your needs here, but you might think about an AI / ML approach here rather than conventional algorithms. See this example from Paolo: [R&D] Machine Learning + Intuiface prototype

Seb

3 Likes

Love this. Paolo for the win!

If its QR codes you need to recognise then use this example that I posted here: Can you do something like this - #15 by Louie_Smith

I hope this helps, either Paolo or this solution should help you.

Louie

2 Likes

Thanks, Louie. I really appreciate your input as always. I think this example is clean and efficient. I will keep you posted on where we go with this.

1 Like

Honestly, sounds over engineered. Make it a single Project or something with Network Triggers, then Excel the combinations as #s that tie to the colors. Wouldn’t be that difficult… wouldn’t mess with AI/Webcam stuff as that’s going to all be buggy and not a great deployment imo

1 Like

I’m all ears, and I’m all for simplicity. Elucidate me. I’d like to hear your insight on this process in detail.