Forums

The 2nd Large-scale Video Object Segmentation Challenge - Track 1: Video Object Segmentation Forum

Go back to competition Back to thread list Post in this thread

> all_frames.zip lacking first-frame annotations

Hi, on downloading the valid_all_frames.zip and test_all_frames.zip I found that they are lacking first frame annotations. Are these datasets supposed to lack first frame annotations or is it a mistake?

Posted by: skwukong @ Aug. 18, 2019, 7:13 p.m.

Hi,
The first frame annotations are included in the valid.zip and test.zip. The all frames zip file only contains image frames. Thanks.

Best,
YouTubeVOS Team

Posted by: linjie @ Aug. 18, 2019, 9:02 p.m.

So, is it that the segmentation masks for the first frame won't be provided and needs to be inferred for "*_all_frames.zip" or will they be provided later on?

Posted by: skwukong @ Aug. 19, 2019, 3:25 a.m.

Hi,
The first frames of "*all_frames.zip" exactly match with those of the valid/test.zip, so you can directly use the first frame annotations in valid/test.zip.

Thanks,
YouTubeVOS Team

Posted by: linjie @ Aug. 20, 2019, 4:49 p.m.
Post in this thread