A Study on Image Privacy Protection in Response to Artificial Intelligence Technology
- Publication Type:
- Thesis
- Issue Date:
- 2023
Open Access
Copyright Clearance Process
- Recently Added
- In Progress
- Open Access
This item is open access.
This thesis addresses image privacy in the era of data sharing and artificial intelligence (AI). Motivated by AI-induced privacy risks, it aims to develop a practical framework and assess its methods. Challenges include inadequate defence against AI, limited privacy definitions, and the need for a privacy-utility balance.
Contributions include a social media image privacy framework using adversarial perturbations, preserving utility while hiding private information. A differentially private image (DP-Image) framework perturbs feature vectors, ensuring privacy against human and AI adversaries. User centric privacy protection allows customizable levels, empowering users in computer vision applications.
The thesis tackles high-quality image de-identification, enhancing anonymized datasets with decoder and attribute optimization techniques. Overall, it significantly advances social media privacy, provable privacy protection, user-centric privacy, and high-quality image de-identification in the age of data sharing and AI.
Please use this identifier to cite or link to this item: