LoSealL / VideoSuperResolution

A collection of state-of-the-art video or single-image super-resolution architectures, reimplemented in tensorflow.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

vespcn pretrained weights

pinkpudding opened this issue · comments

Hi

I have tried the eval.py using
python3 eval.py vespcn -t vid4 --pretrain=../Results/vespcn/save

downloaded both vid4 and vespcn through your google drive url.

However after i have run the command, the results in the ../Results/vespcn/VID4/original seems to be abit off. Understand that the vespcn pretrained model shld be in .ckpt however after extraction i've the following files

  1. checkpoint
  2. vespcn-sc4-ep1000.ckpt.data-00000-of-00001
  3. vespcn-sc4-ep1000.ckpt.index
  4. vespcn-sc4-ep1000.ckpt.meta

is my command and the extracted files correct?

python eval.py vespcn -t vid4 --pretrain ../Results/vespcn/save
Should work now. Make sure in datasets.yaml that VID4 stands for the original path of VID4 dataset:

VID4
--original
-----city
--------frame001.png
--------...
-----walk
--------frame001.png
--------...
-----...

Or it can be:
python eval.py vespcn -t <somewhere>/vid4/input --video --pretrain ../Results/vespcn/save