The training set is newly split from the original val set after rebuttal, as per the suggestion of reviewers. Thus there are not results of OV2Seg trained on LV-VIS.
We are working on video-based training methods now and will include them in our next draft.
The training set is newly split from the original val set after rebuttal, as per the suggestion of reviewers. Thus there are not results of OV2Seg trained on LV-VIS. We are working on video-based training methods now and will include them in our next draft.