Some modifications were made in the new version of eval.py, but removing the logger will cause an error in some environments. MisconfigurationException : Cannot use LearningRateMonitor callback with Trainer that has no logger. Another suggestion: modify the configuration file as follows to avoid warning log outputs: optimize_SMPL:
enable: true
is_refine: true
It seems that your method in the paper also needs to be set to true. I'm not sure why the demo configuration file has it set to false.
Some modifications were made in the new version of eval.py, but removing the logger will cause an error in some environments. MisconfigurationException : Cannot use
LearningRateMonitor
callback withTrainer
that has no logger. Another suggestion: modify the configuration file as follows to avoid warning log outputs: optimize_SMPL: enable: true is_refine: true It seems that your method in the paper also needs to be set to true. I'm not sure why the demo configuration file has it set to false.