Giters

zzh-tech / ESTRNN

[ECCV2020 Spotlight] Efficient Spatio-Temporal Recurrent Neural Network for Video Deblurring

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Using pre-trained model to inference , the results is strange

wangwyForNewTime opened this issue · comments

commented

I choose 15 pictures from BSD dataset to test
I use the comman：
python inference.py --src ./blur/ --dst ./results/ --ckpt ./checkpoints/ESTRNN_C80B15_BSD_2ms16ms.tar

the result is so strange
12
13

I prepare a video to test
Use the comman:
python inference.py --src 1.mp4 --dst ./results/2/ --ckpt ./checkpoints/ESTRNN_C80B15_BSD_2ms16ms.tar
there is an error:
14

Anything I did wrong?

commented

Please share the test data with me. I'll check it for you later.

commented

Please share the test data with me. I'll check it for you later.

thank you
BaiduNetdiskExtractCode：1024
There is my test data,I tried it on both windows and linux ,the results are same

commented

I tested the case with the BSD image sequence as input. There is no problem. The issue may be on your side.

Regarding the video, it seems that the actual number of frames does not match the number read by cv2. Anyway, I've updated the code that extracts frames from the video to avoid this.

commented

I tested the case with the BSD image sequence as input. There is no problem. The issue may be on your side.

Regarding the video, it seems that the actual number of frames does not match the number read by cv2. Anyway, I've updated the code that extracts frames from the video to avoid this.

Thank you for your reply
There is no error reminder when test BSD images ，I think the strange result maybe because of the different version of environment library version
Accroding to ReadMe:
11
I use Annaconda to create a new environment,there is conda list:
12

It could be the version，can you give me the specific version which you are using?

commented

The list of dependencies of my current env (generated by pipreqs) is as follows for your reference:

lmdb==1.3.0
numpy==1.21.5
opencv_python==4.5.3.56
Pillow==9.2.0
rawpy==0.17.1
scikit_image==0.19.2
thop==0.1.0.post2207010342
torch==1.11.0
torchvision==0.12.0
tqdm==4.64.0

commented

The list of dependencies of my current env (generated by pipreqs) is as follows for your reference:

lmdb==1.3.0 numpy==1.21.5 opencv_python==4.5.3.56 Pillow==9.2.0 rawpy==0.17.1 scikit_image==0.19.2 thop==0.1.0.post2207010342 torch==1.11.0 torchvision==0.12.0 tqdm==4.64.0

Thank you,the problem is solved.
This is really version problem.Now coding runs smoothly
In addition to the above, also need python 3.7