Open schuyler-cohen-zipline opened 1 year ago
fyi @Christopher22 I believe you added this if you can verify this? I don't have a great way to test/prove its correct other than the name/looking through the c code at how it's used.
The apriltag source code sets cos_critical_rad
but does not make any use of critical_rad
. It makes sense because the valid angle is 180 degrees at maximum. I prefer to let the user set cosine value directory like what apriltag.c
does, and cache the acos(min_angle)
.
The apriltag source code sets
cos_critical_rad
but does not make any use ofcritical_rad
. It makes sense because the valid angle is 180 degrees at maximum. I prefer to let the user set cosine value directory like whatapriltag.c
does, and cache theacos(min_angle)
.
I think from an api perspective, setting an angle is much more intuitive than trying to decide what an appropriate "cos of an angle" to pass a function is, but I agree with you, the C version doesn't use it, and it almost seems like it's not really a fully formed tunable parameter in the C version. I've updated it to take the cos value, what do you think?
We can add a checked set_min_angle()
to get the best of both.
In QuadThresholds, the parameter min_opposite_angle was being used to directly set cos_critical_rad, but that's just a cached version of cos(critical_rad), not an entierly different parameter.
This sets it to the proper value.