The project is an official implementation of our CVPR2019 paper "Deep High-Resolution Representation Learning for Human Pose Estimation"
deep-learning
human-pose-estimation
mpii-dataset
mpii
mscoco-keypoint
coco-keypoints-detection
deep-high-resolution-net
high-resolution-net
-
Updated
Dec 17, 2020 - Cuda
Hi everybody
Currently only one sample per GPU is possible while executing tests, which leads to very slow validation and testing for me. I assume to fix this the
forward_testmethods for theTopDownandBottomUpclasses would need to be updated. Is this planned anytime soon or should I give it a try myself to contribute?