embracing / active3dpose Goto Github PK

Thanks for your great work!!!
(1) Can you provide new checkpoints download link to evaluate your agent?
(2) What's your training environment setting for this algorithm, just like how long does this training process become converged on which gpu and cpu settings?
(3) What's more, all the four download binary environments are ten women in the blank environments, it seems like that other three environments (gym, street, grass) didn't upload yet

https://disk.pku.edu.cn/#/link/56F3EA16170527EABFA1F0497B66D1F2
Valid Until: 2023-05-05 23:59

You can find the model checkpoints of the control policies for 3 to 5 cams in the zip file. I did an evaluation on the Blank Environment with each model myself to save you the troubles. You can view the evaluated results with our PyQT visualizer.

Originally posted by @mickelliu in #1 (comment)

How to use custom camera input and process them instead of using the simulation output of unreal engine?
Please let me know about it.
Thank you

Thanks for your great work!

I tried to run exec_env.py under the Active3DPose-Main but it shows ‘no module named activepose’

Thanks for your great work!

Is offering a policy checkpoint for the following evaluation possible?

python run/scripts/evaluate.py --ckpt ray_results/mappo_ctcr_wdl/PPO_1bd01_00000_Feb08