vt-vl-lab / FGVC

[ECCV 2020] Flow-edge Guided Video Completion
Other
1.55k stars 262 forks source link

saved images contain green masks #16

Closed WeidiXie closed 3 years ago

WeidiXie commented 3 years ago

Hi,

May I ask, why this line exists ? https://github.com/vt-vl-lab/FGVC/blob/9820d3c1a33ba402009ecb1d25e897cbcddc74d5/tool/video_completion.py#L373

and this line https://github.com/vt-vl-lab/FGVC/blob/9820d3c1a33ba402009ecb1d25e897cbcddc74d5/tool/video_completion.py#L548

These will make the saved images to have green masks on top.

andrewdevcup commented 3 years ago

Looks like those are leftovers from developers, just mark them as comments. Maybe they did it for the people who wants the mask only, and didn't remove it.

WeidiXie commented 3 years ago

@Andrew-afk
I see, thanks a lot !

gaochen315 commented 3 years ago

Andrew is right, this is only for debugging purpose. The green indicates the regions that are not filled after the current iteration. You should not see any green region in the final result, i.e., frame_comp_xxx folder with the largest xxx.

gaochen315 commented 3 years ago

Close the issue as it was solved.