Slashdot Mirror
Video Inpainting Software Deletes People From HD Video Footage
cylonlover writes "In a development sure to send conspiracy theorists into a tizzy, researchers at the Max Planck Institute for Informatics (MPII) have developed video inpainting software that can effectively delete people or objects from high-definition footage. The software analyzes each video frame and calculates what pixels should replace a moving area that has been marked for removal. In a world first, the software can compensate for multiple people overlapped by the unwanted element, even if they are walking towards (or away from) the camera."
Don't the new Galaxy S4 have a similar feature, if I read correctly? Although only for photos.
/ The Arrow
"How lovely you are. So lovely in my straightjacket..." - Nny
I liked the fact that you could still see the pedestrians in the reflections of the display window in the video of the musicians, even though they had been erased from the front end. Like the vampire test, but the other way around.
If we extrapolate this, perhaps we won't be able to trust video as evidence any longer, so there's no reason to have all these surveillance cameras around.
What you proposed isn't that far-fetched, as I ended up having to contrive and implement the equivalent of this, i.e., passive, automated estimation of body shape under clothing, either from a single image or from multiple video frames, for some work I did in action recognition that required a fairly accurate representation of the person's proportions. Others, e.g., A. O. Balan and M. J. Black, "The naked truth: Estimating body shape under clothing," in Proceedings of the European Conference on Computer Vision (ECCV), 2008, pp. 15–29, have come up with solutions too.
Rainbows End, by Vernor Vinge. What you described is just the beginning.
(-1: Post disagrees with my already-settled worldview) is not a valid mod option.
...The idea of element extraction forms the basis of various camera enhanced video games found on the current consoles. Usually, the technique is the reverse of the example in the article, where it is the background that is removed so that the player may be isolated and inserted into a virtual scene.
Uh, yeah, and now that we've revealed that removing someone is "ancient technology" is is exactly this reversed scenario that most should fear today and is ripe for abuse in a corrupt world.
One day, you think it's cool that you've been "painted" into a video game...until you realize that same technology can "paint" you right into Exhibit A: The murder scene.
How long before innocent people are framed? Judges can't even understand how the internet works. You think they're going to grasp this and give you a fair trial?
Yes, it does, as the video demonstrates. Their algorithm requires a static background (i.e. a stationary camera) and handles and moving foreground objects as user-selected special cases. If the background were to move, they'd need to either motion-compensate the entire footage first (assuming the camera only changed orientation, and not position, so parallax was not an issue), or perform an exhaustive search over the entire footage (which is the specific situation their algorithm is trying to avoid!).
Anyone reminded of the Laughing Man from Ghost in the Shell: Stand Alone Complex?
A hacker who was able to hack the cybernetic vision of others in real-time to make himself invisible...