yulin010101
yulin010101
I change the code in detectron2's built.py to let the milestone less than total number of updates. You can try that. It can run but I am not sure if...
How do you get the example image Will you get better training results if you turn off mosaic?
Can you explain more?
It looks like all data was detected as corrupted. Try to restore your data first.
I see. Thank you for your explanation.
For pt file you could refer to #901. For freezing you can experiment to see the effect and --freeze 50 should work.
What adjustments did you make?
Did you set the training epoch to be greater than fifty?