Inquiry Regarding the Ranking Rules for Decoding Time

[gorgeous]

Dear Organizing Committee,

I would like to inquire about the specific rules concerning the restriction that the slowest 25% in terms of decoding time are not eligible for awards.

1. For a team’s final decoding time, is it calculated as the average decoding time of the three bitrate points from the team’s last submission on the leaderboard during the testing phase? Or is it the average decoding time of all submissions from that team during the testing phase? Alternatively, is it determined by the longest single decoding time?

2. Regarding the ranking of decoding times, are all submissions from all teams on the leaderboard during the testing phase considered, or is only the last submission from each team used for ranking? Additionally, will baseline methods (e.g., HM) be included in the decoding time ranking?

Best regards,

Thanos

[Jona Ballé]

Dear Thanos,

1. For the image track, what matters will be the average decoding time across the bitrates.

2. In general, we care about the decoding time of the submissions that we run the human evaluation on. So if you do multiple submissions during the test phase, and the earlier ones are replaced by newer ones, then the earlier ones won't matter.

The baselines do not count in terms of decoding time or prizes, but we may still include them in the human eval.

Hope this helps!

~ Jona

--
You received this message because you are subscribed to the Google Groups "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 visit https://groups.google.com/d/msgid/clic-list/166ed015-53ed-4723-83b8-e062a294a70cn%40googlegroups.com.