You are viewing the course site for a past offering of this course. The current offering may be found here.
Lecture 21: Image Sensors (10)

In OH, Professor Ng mentioned how image sensors could (potentially) be a solution to the issue of ubiquitous imaging and everyone's photo being on the internet. If everyone had a way to set their "privacy mode" to on (like on a government website), then at a low level, image sensors in all cameras could somehow filter out the faces of people who don't want to appear in images!


This is an awesome idea! I assume then it might requires a combination of computer vision (for face recognition) and graphics (for masking private faces).


That sounds like a cool idea, though with that, I imagine there would have to be some sort of database identifying people in order to identify those who want to stay private. With that, I think there comes another security risk in protecting that database. I wonder how it can be best executed.


Yes, I agree it was a really interesting point during lecture that since we start to see this trend of ubiquitous imaging we can start to take precaution from a lower level such as image sensor. I was wondering if you could even do something where the image sensor itself can cryptographically sign every image it captures and detects with some sort of privacy tag that can be embedded within the data of the image. That way whenever that image is trying to be shared, the tag will have to match some sort of protocol that makes it valid.

You must be enrolled in the course to comment