The order of config files execution
ShuoYang-1998 opened this issue · 2 comments
Hi, Thanks for your excellent work!
I found the config files in ./configs/OWOD/ folder and the data split files in ./datasets/OWOD_imagesets/ are extremely confusing. What does t#_train_with_unk.yaml mean? What does t#_ft.yaml mean? Should I have to run t#_val.yaml before t#_test.yaml? Can you provide a config file execution order if I want to reproduce the ORE results? E.g., t1_train.yaml->t1_val.yaml->t1_test.yaml->t2_train.yaml ......
It would be very appreciated if you can release several run.sh that consists of all training schedules in the paper.
Thanks!
What does t#_train_with_unk.yaml mean?
You can ignore this. This is to train with unknown labelled so. This is used to report the 'Oracle' result in Table 2.
t1_train.yaml->t1_val.yaml->t1_test.yaml-
Thats correct.
run.sh
I will definitely do. Please allow some time.