3.3 detect over video

[tjulitianyi1997]

Can someone implement the function of the author's video on Youtobe website? When I run the following sentence, I keep showing no face.

python infer_on_video.py -f [video file name] -s [save file name]

Has anyone encountered the same situation? Is there any solution?

[sunweixings]

I meet the question the same with you.And the file named recording can not open.Have you solved this question?

[zhangbc12]

I can not detete the face too. Have you solved this problem?

[charmere]

the video file is on data/facebank/ which is Default location. so you can set your video at data/facebank/video_name.

[Shame-fight]

I have the same question.always no face ,who can give me some advices?

[mzm1183710118]

@Shame-fight me too，have you solved this problem？

[thunanguyen]

I think it is the problem with pytorch version. Switch to new implementation of mtcnn-pytorch is fine for me but still cannot open the output video, I think it's a problem of opencv version

Update: I've found out why the output video can not open because the video writer use fixed and (possibly) incorrect frame height and width, if you change to the actual height and width of your video frame then it works