Chenyang LEI comments

Results 20 comments of


                                            Chenyang LEI

Dataset-2 cannot be opened after downloading.

Can you show the error message? We check it again. And we can download it and extract it

[Question] - How to improve the color in deflickered video?

Can you try the following two things: (1) Upload the results/YOUR_VIDEO_NAME/neural_filter/concat.mp4 (2) Try to use a small number of frames to see the performance, e.g., 100 frames.

[Question] - How to improve the color in deflickered video?

Interesting results. Then I think that is because the structures of objects change a lot (deformable distortion) and the motion is very large. In this case, the extracted atlas is...

"CUDA out of memory" for large video file

Hi, The preprocess optical flow cannot support high-resolutin images : ( We will try to find a more efficient preprocessing optical flow estimation model. Another way is to use the...

[BUG] Using image folder path - Expected to have 3 channels, but got 4

Yes, I think you are correct. It is because your images are PNG and there are 4 channels for your PNG images. Let us fix this bug.

[BUG] Using image folder path - Expected to have 3 channels, but got 4

> Excuse me when can fix this bug, I also encountered this problem Can you share your video so that we can test it? Thanks.

The intrinsic of dataset

Please check the vd_local.npy

The intrinsic of dataset

I need time to check the details. It is an old paper.

Code about the warping error metric

Can you check the reply here? https://github.com/ChenyangLEI/deep-video-prior/issues/4#issuecomment-730376178 I am quite busy these days with CVPR.

where is the pytorch vision？

https://github.com/ChenyangLEI/flash-reflection-removal/tree/main/torchfrr