Closed MandyDongrs closed 1 year ago
应该是跟mmrotate的dota处理一样,三个json的内容除了路径其他都是一样的吧。
just replace the key mode
from train
to val
and run the split script, you will get the sub-images and corresponding annotations about val-set, likewise for test-set.
Model/Dataset/Scheduler description
Under data/sodaa/split_configs directory, I only find sodda_train.json. Is the configuration of val and test the same as train ( except for the path settings) ? Or could you provide the sodaa_val.json and sodaa_test.json files, as you mentioned in the data structure of sodaa/readme?
Open source status
Provide useful links for the implementation
No response