-
Notifications
You must be signed in to change notification settings - Fork 42
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
PCKh/head size calculation #14
Comments
Hi Fabian, Thanks for reporting this issue. You are right, the calculation in this repo is not PCKh. These functions were taken from the official repository and I hadn't realized their note before: https:/princeton-vl/pose-hg-train#training-accuracy-metric So the numbers in the Table A.1 of the supplemental material correspond directly to the output of our released code. Best, |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Hi Gül,
according to your BodyNet paper, you're using [email protected] metric for evaluating the performance of the 2D pose estimation. Newell et al. refer to PCKh as a "distance [that] is normalized by a fraction of the head size."
Is the calculation of PCKh included in this repo? The only PCK related calculations I found are in eval.lua L14 and L93. While L14 seems to be a generic implementation, the call in L93 seems to calculate the [email protected] metric assuming that the image/heatmap fits the person.
How did you calculate the head size/PCKh for the evaluation in your paper?
Thanks in advance.
Best,
Fabian
The text was updated successfully, but these errors were encountered: