search

CAMMA-public / SelfPose3d

Official code for "SelfPose3d: Self-Supervised Multi-Person Multi-View 3d Pose Estimation"
Other
24 stars 2 forks source link

README not reproducable: CMU Panoptic dataset image extraction missing #3

Open LadnerJonas opened 3 months ago

LadnerJonas commented 3 months ago

Recently, I received access to a cuda-machine to test this project. I am now trying to continue reproducing the README. Afterward, we want to run SelfPose3d on our videos and data.

To my current understanding, SelfPose3d requires to extract images / hdPose3d / calibration from the video data set. This can easily be done using the panoptic-toolbox script. e.g. ./panoptic-toolbox/scripts/extractAll.sh SelfPose3d/data/panoptic-toolbox/data/160224_haggling1/

After that, all necessary folders and files are generated.

Can you please add something similar to the project README file?

LadnerJonas commented 3 months ago

Also, please note that the 16060224_haggling1 is now called 160224_haggling1

keqizero commented 3 months ago

Hi,

Yes, we extracted the images by following the scripts written in the Panoptic repository. Thank you, and I will update the README file.