-
Notifications
You must be signed in to change notification settings - Fork 336
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add configurable max_hamming_distance for the AprilTag Detector #93
Add configurable max_hamming_distance for the AprilTag Detector #93
Conversation
@wxmerkt Can we get this merged if its good to go? |
4793e17
to
e00abe9
Compare
Based on upstream code comment: > Tunable, but really, 2 is a good choice. Values of >=3 > yonsume prohibitively large amounts of memory, and otherwise > you want the largest value possible.
I added comments on the parameter setting from upstream which are now passing through the buildserver. I am not certain if there isnt a loss of quality of detection reducing the setting? How much was the memory usage before/after? |
The memory usage drops from about ~6 GB for the 52h13 family to ~ 200 MB.
So memory allocated = (48714 + 48714 * 52 + 48714 * 52 * 51) * 3 * 12 = 4743769320 bytes ~ 4.73 gigabytes with max_hamming_distance = 2. For max_hamming_distance = 1 -> If we reduce the max_hamming_distance it should increase the accuracy as we are allowing fewer bit errors to be looked up. However some valid detections might dropped because we are allowing fewer errors. Irrespective I think being able to configure the parameter is valuable and we retain the original functionality by setting 2 as the default parameter. |
Thank you for the detailed explanation and contribution. :-) |
From Pull request AprilRobotics#93
Adds configurable max hamming distance bits for the AprilTag Detector.
Context:
I might have missed some documentation files which might need to be updated. Happy to take a look at those.