Hi, SeungjunNah
Could you please share how do you get this results? don't need to be detailed, like a one stage method or two stage method something...
Because If there is a baseline can get such a score, I think I need to change direction....
Thanks so much.
Hi zhangqi,
It is not from a specific algorithm. I just upsampled the corresponding LR images via bicubic interpolation.
The performance might look too good to be a baseline, but it's because I used odd LR frames.
I uploaded the dummy submission just to verify that the CodaLab evaluation server works.
You don't need to consider it a baseline.
An appropriate baseline would be overlaying or interpolating the even-numbered frames.
Hi zhangqi,
Could you tell me the result '01/26/21 30.89 (1) 0.8650 (1)' you submitted is consistent with rule of training/test data usage?
Thanks so much.
Hi,zhuyeye
Sorry to reply late.
No, my submit is just for test, I use all the frames.
I also want to know these three submits is legal or not? Do you know?
3iPrayer 02/20/21 29.54 (3) 0.8315 (2) 1.740 (11) 0.0 (2) 0.0 (2)
4 swarnendu 7 02/17/21 28.89 (4) 0.8125 (3) -1.000 (1) -1.0 (3) -1.0 (1)
5 whu_hj 1 01/08/21 26.30 (5) 0.7191 (5) 0.009 (2) 1.0 (1) 0.0 (2)
Dear zhangqi,
At the moment, we don't have any information if the current validation leaderboard submissions use the even- or odd-numbered frames.
I expect the final submissions with the fact sheets (updated on Jan 29, 2021) will clarify them.
Also, please note that by submitting the fact sheet, the participants claim they have followed the challenge rules.
The scores in the development phase do not affect the final ranks.
Best,
Seungjun