how to gen unlabeled.json in my dataset

[HeuristicLU]

DO you know how to gen unlabeled.json ? i use my custom dataset, thanks

[luisfra19]

I am also using my own dataset; I followed the data partition described on readme with a small change with the symbolic link. I get partitioned data labeled and unlabeled, however the unlabeled data still has annotations. So I am not sure if it is correct or not.

[xiangtaowong]

I am also using my own dataset; I followed the data partition described on readme with a small change with the symbolic link. I get partitioned data labeled and unlabeled, however the unlabeled data still has annotations. So I am not sure if it is correct or not.

maybe u can read the raw format of unlabel_annotation.json in coco, and write a script, like {"images":[ {"file_name": "", "height": , "width": , "id": }, ...... ], "annotations": [], "categories": []}