You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hey there, awesome implementation. I am trying to run the code to see how it performs, however I have a quick question on the input folder structure. In the code there is a check if the input image name is 0 and a check if it is 1. Which of these values is intended as the starting value of the video folder?
The text was updated successfully, but these errors were encountered:
Hi @stefanklut ,
The implementation uses glob to grab all pngs in a folder. So no need to worry about the indexing of the first frame I think.
By the way, we use ffmpeg to extract video into frames. ffmpeg -i xxx.mp4 %05d.png
The first frame will be named 00001.png.
Thank you, the ffmpeg info was just what I needed. However, I do think filename matters since at line 235 on run_FuSta.py there is a check if the filename is 0.
That number is related to the smoothed homography matrices and flows generate by CVPR2020CODE_yulunliu_modified. The if == 0 part can be ignored I think.
Hey there, awesome implementation. I am trying to run the code to see how it performs, however I have a quick question on the input folder structure. In the code there is a check if the input image name is 0 and a check if it is 1. Which of these values is intended as the starting value of the video folder?
The text was updated successfully, but these errors were encountered: