An Efficient Approach to De-Identifying Faces in Videos
What is it about?
A novel approach to face de-identification for privacy protection, which is able to 1. conceal original identity of the faces in a video sequence; while 2. retain the original expressions on the faces after de-identification; and 3. operate in real-time for video streaming at 30 frames/sec
Why is it important?
Our novel approach has been tested extensively with real-life video sequences. Results show that our approach has achieved: privacy protection, identity consistency, preservation of facial expressions, and real-time operation.
The following have contributed to this page: Li Meng
In partnership with: