search

xinntao / EDVR

Winning Solution in NTIRE19 Challenges on Video Restoration and Enhancement (CVPR19 Workshops) - Video Restoration with Enhanced Deformable Convolutional Networks. EDVR has been merged into BasicSR and this repo is a mirror of BasicSR.
https://github.com/xinntao/BasicSR
1.48k stars 320 forks source link

what is woTSA? #214

Open gulzainali98 opened 2 years ago

gulzainali98 commented 2 years ago

what is woTSA and which configuration is used in the paper L or M?

ckkelvinchan commented 2 years ago

woTSA means the model without TSA fusion. In the paper, we report the results of the "L" model, except for Table 4.

gulzainali98 commented 2 years ago

So first you train model and then you train only TSA module with a pre trained model? Am I correct?

gulzainali98 commented 2 years ago

can the same configurations in options/train/EDVR/ be used for video deblurring as well?

gulzainali98 commented 2 years ago

Also can you please tell me what is the logic behind calculating total number of iterations per epoch and calculating epochs based on that? Why number of epochs go down when I reduce the batch size? Thank you very much. Looking forward to your response :)