This is an official implementation of our CVPR 2020 paper "HigherHRNet: Scale-Aware Representation Learning for Bottom-Up Human Pose Estimation" (https://arxiv.org/abs/1908.10357)
Hi, I tested a demo code from pull-request on some images, and found that there are a lot of overlapping outputs.
I think this is the problem of nms in your code.
Do you have any idea to do better nms for noisy, redundant predictions?
And how did you pick the result for benchmark evaluation?
Hi, I tested a demo code from pull-request on some images, and found that there are a lot of overlapping outputs. I think this is the problem of nms in your code.
Do you have any idea to do better nms for noisy, redundant predictions? And how did you pick the result for benchmark evaluation?