The sequence resolution problem in validation set

[li ming]

Dear Organizer.

          We note that the Y component resolution is twice as high as the other two components(U and V), which is mentioned in the video compression track. However, the Y component resolution of the sequence Lecture_1080P-4991 in validation set is 959x720 and the resolution of U and V components is 480x360，which doesn't satisfy the relationship mentioned above.

          We wonder whether there is a problem in the selection of the sequence lecture_1080P-4991? Can it be replaced or deleted?

Best regards

gang

[Nick Johnston]

Adding Erikur and Fabian who worked on the dataset release. I think we may have ran into something similar last year.

--
You received this message because you are subscribed to the Google Groups "Workshop and Challenge on Learned Image Compression (CLIC)" group.
To unsubscribe from this group and stop receiving emails from it, send an email to clic-list+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/clic-list/d8103571-9d6d-402c-8a06-608b82a8227bn%40googlegroups.com.

--

=====================
Nick Johnston
Software Engineer
Machine Intelligence, Compression

[Fabien Racape]

Hi Nick, 

Following up on this. Have you decided whether this sequence will be left as is or replaced by a 960x720 version?

Thank you, 

Fabien 

[Lucas Theis]

Hi Fabien,

I have not yet heard the other organizers' opinion on this specific request, but the dataset is very unlikely to change now. I recommend you use padding and cropping to handle a wider range of resolutions.

To view this discussion on the web visit https://groups.google.com/d/msgid/clic-list/52cd5785-a2e9-4481-b858-beb338f6fac0n%40googlegroups.com.

[Fabien Racape]

Thanks for your reply, Lucas.

-Fabien