Closed yupei-git closed 3 years ago
That is solved. :)
Hello @yupei-git , can you tell me how you solved this ?
I reinstalled cosypose and the environment. Then copied the local_data from the original cosypose into the new one. When I ran the train again, the above error did not exist.
Hello, when I got to the 'Training details' in the 'BOP20 models and results' section, using
'runjob --ngpus=1 python -m cosypose.scripts.run_detector_training --config bop-ycbv-synt+real'
to train the ycbv dataset on the synthetic + real image, the problem occursI modified the job-runner-config.yaml, as shows:
and used '
runjob-config job-runner-config.yaml
', but got the same error.Before that, the 'Installation' and 'Downloading and preparing data' sections were successfully completed. Since I wanted to do it with a single local GPU, I just used the command line '
export CUDA_VISIBLE_DEVICES=0
' and 'python -m cosypose.scripts.example_multigpu
' in the 'Note on GPU parallization' section. But that works in 'Reproducing single-view results' part.Could anyone help me with this? Thanks in advance!