Where are the unknowns for the first task coming from ?

[manoja328]

There are lots of text files in the https://github.com/JosephKJ/OWOD/tree/master/datasets/OWOD_imagesets folder.

Am i supposed to use t1_train.txt ... and so on for training , all_test for testing , all_task_val for validation?
if so, the first test image sets ( t1_train.txt) all has images/classes from VOC dataset only . So how are the unknowns metric WI and AOSE calculated?

Related, what are the files [t1_train_with_unk.txt] .. etc. used for?

I assume _ft files are used for the finetune model?

Thanks.