This repository is an official PyTorch implementation of the paper "Learnable Triangulation of Human Pose" (ICCV 2019, oral). Proposed method archives state-of-the-art results in multi-view 3D human pose estimation!
MIT License
1.09k
stars
181
forks
source link
Video Quality Issue in CMU Panoptic Benchmark #147
Hi, thanks a lot for your great job! I'm trying to train your model on CMU Panoptic benchmark. I found that there are a few videos with blurred images, offset images (such as 171204_pose5/hdVideos/hd_00_16 frame 25108/25110 etc). I'm wondering if you just ignored such frames?
Hi, thanks a lot for your great job! I'm trying to train your model on CMU Panoptic benchmark. I found that there are a few videos with blurred images, offset images (such as 171204_pose5/hdVideos/hd_00_16 frame 25108/25110 etc). I'm wondering if you just ignored such frames?