5
$\begingroup$

I know some techniques for augmenting data when images are used, but I don't know if there are any such techniques specifically catered to videos. Since videos have an additional temporal dimension compared to images I was wondering if there were any techniques that augments the video data using the view points and the motion information etc

$\endgroup$
3
$\begingroup$

You can augment videos in the temporal dimension through clipping, or taking random sequences of consecutive frames. You can also augment in the spatial dimension by cropping frames randomly to simulate camera jitter. Unfortunately, augmenting video data by extrapolating motion information or by producing a video taken from a different viewpoint seems extremely computationally intensive.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.