Closed Undercut closed 4 years ago
Still under Results/model-name/infer-folder-name
Still under
Results/model-name/infer-folder-name
谢谢回答,我这里还有一个问题想请教,我在测试使用Vespcn-tensorflow获得超分辨率视频,15秒24帧720P的视频被我切分成200多张图片,这组图片序列通过Vespcn需要非常长的时间,大约1个多小时(在2600X上进行测试,未使用GPU加速),请问有没有什么办法能够加快预测的速度,使用Pytorch会有帮助吗? 同时我查阅了Vespcn那篇论文,原论文声称使用了动作补偿来加快视频超分辨率化的速度,但在这个模型上似乎并没有体现,请问是否有完整复现论文的模型?
Thank you, I got it. By the way, which model should i choose if i want a fast VSR.
FRVSR is highly recommended
Thank you so much.
hi,when I test the vespcn-tensorflow, it can't save image to infer,can you give me some advice ? @Undercut @LoSealL
It just return
Test: 0it [00:00, ?it/s]
my command is
python eval.py dbpn -t vid4 --pretrain=../Results/vespcn/save
You should use
python eval.py vespcn -t vid4 --pretrain=../Results/vespcn/save
Thanks, I see that you have change the '/VSR/Backend/TF/Framework/Trainer.py', it works for testing vespcn model. Thank you very much!
Hi, when I use python eval.py vespcn -t vid4 --pretrain=../Results/vespcn/save
The terminal reports some issues. I don't know how to solve it. Can you give me some advice?
Here is the report information:
(TF2.1) D:\GP_AI\Super Resolution\VideoSuperResolution\Train>python eval.py vespcn -t vid4 --pretrain=../Results/vespcn/save
2022-05-07 20:01:40,213 INFO: LICENSE: VESPCN is proposed at CVPR2017 by Twitter. Implemented by myself @LoSealL.
Traceback (most recent call last):
File "eval.py", line 122, in
I found the result images in the Results/model-name/test-datasets-name. But if i use <--infer >,where to find my result images?