Hey all,
I was able to runa customized training on my own data with cosypose with an Quadro P4000 graphic card. But to get some usefull results i have to train several days... So i decided to upgrade to a RTX 3080ti graphic card, BUT I had to update the nvidia driver (440 -> 460/470) and because of this also CUDA (10.2 -> 11.2) ... also i had to update pytorch, cudnn, pybullet ...
After some problems running cosypose (run_pose_training.py) I have another problem:
In caching.py (row 45) is a "deepcopy", which gives me an error:
module 'pybullet' has no attribute '__deepcopy__'
File "/home/scale/dev_new/cosypose/cosypose/simulator/client.py", line 10, in __getattr__ (Current frame)
attribute = getattr(pb, name)
File "/home/scale/anaconda3/envs/cosypose_develop_new/lib/python3.9/copy.py", line 151, in deepcopy
copier = getattr(x, "__deepcopy__", None)
If I change "deepcopy" to "copy", there is no error anymore and the "run_pose_training.py" is starting as before... BUT now there is some memory leak... perhaps because of the "deepcopy->copy" change? I do not know...
Does please anybody give me some hint how to solve these "deepcopy" problem? Does anybody can help me please??
Hey all, I was able to runa customized training on my own data with cosypose with an Quadro P4000 graphic card. But to get some usefull results i have to train several days... So i decided to upgrade to a RTX 3080ti graphic card, BUT I had to update the nvidia driver (440 -> 460/470) and because of this also CUDA (10.2 -> 11.2) ... also i had to update pytorch, cudnn, pybullet ...
After some problems running cosypose (run_pose_training.py) I have another problem:
In caching.py (row 45) is a "deepcopy", which gives me an error:
If I change "deepcopy" to "copy", there is no error anymore and the "run_pose_training.py" is starting as before... BUT now there is some memory leak... perhaps because of the "deepcopy->copy" change? I do not know...
Does please anybody give me some hint how to solve these "deepcopy" problem? Does anybody can help me please??