Theses and Projects: Difference between revisions

no edit summary
No edit summary
No edit summary
Line 30: Line 30:


===  * '''New!''' Privacy-preserved Video Analytics  ===
===  * '''New!''' Privacy-preserved Video Analytics  ===
In this project/theses topic, you will explore how to protect privacy in videos and how to recover them.
This project/thesis topic focuses on the protection of privacy in videos and the development of techniques for privacy-preserving video recovery. The project involves three key tasks:
This project includes 3 tasks: 1) deploy system with yolo and CycleGANs/DataGen (coded available); 2) set blur degree by trading-off between inference accuracy (e.g., Yolo for detection) and privacy protection; 3) Finally, adapt blur degree for Pan-tilt-zoom Cameras.
 
Implementation of a system utilizing YOLO and CycleGANs/DataGen for video analysis and processing. Code for this is already available for use.
 
Development of a privacy protection mechanism by adjusting the level of blur applied to the video, taking into account a trade-off between inference accuracy (e.g. detection by YOLO) and the level of privacy protection.
 
Optimization of the blur level for Pan-tilt-zoom cameras, to ensure that the system is effective at capturing key visual information while still preserving privacy.


Please contact Dr. Tingting Yuan [tingting.yuan@cs.uni-goettingen.de] (B/M/P)
Please contact Dr. Tingting Yuan [tingting.yuan@cs.uni-goettingen.de] (B/M/P)
536

edits