Would love to use

[liamzebedee]

Hi mate!

I stumbled across your project and I'm very excited to use it. I'm finding it a bit difficult to read the code. A couple questions:

1. is chromaTagRealTime using a simpler version of the Chromatag algorithm? Can only seem to see the A channel of LAB being used here and am a bit confused if it offers the same efficiency
2. where are your modifications implemented? I'm looking to use this right now in place of apriltag on an RPI project/assignment.

Cheers - and by the way, absolutely love you writing. Keep it up!

Liam

[lettergram]

Hello!

1. So, this repo is more-or-less a wrapper around AprilTags. I.e. it uses the same algorithm, with a pre-precessing step. This dramatically increases performance. At the moment, (in this repo) only the A channel is used. In a future iteration of the code the B channel was used to encode more info + an additional check (i.e. reduction in errors). Additional work was done by a grad student I was working with: https://arxiv.org/abs/1708.02982 He expanded on this idea, and did more optimization(s).

2. The only difference between ChromaTags (this implementation) and AprilTags the reduction of superfluous lines that need to be computed on. i.e. We are only looking at lines in the A channel (not B or L), thus when we go to the April tags algorithm, we only have to compute the corners for a very small subset of lines in the frame (typically this is referred to as pre-processing). From line 107 to 178 is just AprilTags code; the rest of that file are the modifications I made. Again, further work was done, but honestly I don't know where that code is....

Sorry, wish I could be more helpful