SZChang

Could you support HRNet?

focal loss pls！

TTA both of mmdet and mmseg support.

Thank you!

@Ezra-Yu Hi, sorry to bother you again. If I use `Repeatdataset(times=15)` in `k-fold` of `train_cfg`, should I also use `Repeatdataset(times=15)` in `val_cfg` and `test_cfg`? Like this? ``` train=dict( type='KFoldDataset', dataset=dict(...

And if I really want to use "data/my_data/val" and "data/my_data/meta/val.txt" as test_set, how can I change the cfg？

You mean this? ``` train=dict( type='KFoldDataset', dataset=dict( type='RepeatDataset', times=15, dataset=dict( type=dataset_type, data_prefix='data/my_data/train', pipeline=train_pipeline)), fold=0, num_splits=5), val=dict( type='KFoldDataset', dataset=dict( type='RepeatDataset', times=15, dataset=dict( type=dataset_type, data_prefix='data/my_data/train', pipeline=train_pipeline)), fold=0, num_splits=5), test=dict( type=dataset_type, data_prefix='data/my_data/val', ann_file='data/my_data/meta/val.txt',...

For 1， NO There is an error:  config: ``` train=dict( type='KFoldDataset', dataset=dict( type='RepeatDataset', times=15, dataset=dict( type=dataset_type, data_prefix='data/my_data/train', pipeline=train_pipeline)), fold=0, num_splits=5), val=dict( type='KFoldDataset', dataset=dict( type='RepeatDataset', times=15, dataset=dict( type=dataset_type, data_prefix='data/my_data/train', pipeline=train_pipeline)),...

> For 1， NO There is an error:  > > config: > > ``` > train=dict( > type='KFoldDataset', > dataset=dict( > type='RepeatDataset', > times=15, > dataset=dict( > type=dataset_type, >...

或者说我想在K-fold的情况下，仅在train中使用repeatdataset，然后在val中就不使用了，请问这种该如何办到？