search

HRNet / HigherHRNet-Human-Pose-Estimation

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)
MIT License
1.33k stars 272 forks source link

How to reduce output layers for lower number of joints? #30

Closed H19012 closed 4 years ago

H19012 commented 4 years ago

I want to train with only few joints. How do I reduce the output layers to accommodate that? Currently pred(17) and gt tensors are not matching.

H19012 commented 4 years ago

Ok, found the solution in the yaml file( there are two NUM_JOINTS, one for DATASET and another for MODEL) and also here: https://github.com/HRNet/HigherHRNet-Human-Pose-Estimation/blob/master/lib/models/pose_higher_hrnet.py#L311