Pose-Group / DCPose

This is an official implementation of our CVPR 2021 paper "Deep Dual Consecutive Network for Human Pose Estimation" (https://openaccess.thecvf.com/content/CVPR2021/papers/Liu_Deep_Dual_Consecutive_Network_for_Human_Pose_Estimation_CVPR_2021_paper.pdf)

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

performance

chenrxi opened this issue · comments

commented

I ran training new model on posetrack17 and test it on validation dataset for three times. I only got 80.9mAP for three times where there are large gap. 2 * 2080 and other config is not changed. I used 0, 1000, 8000 random seed for different initialization. So where's the problem.

Excuse me, can we only use resnet for training? It seems that there is no run file of hrnet?