Video-DPRP: A Differentially Private Approach for Visual Privacy-Preserving Video Human Activity Recognition

2025-03-03

AI Summary

Existing video-based human action recognition (HAR) methods struggle to jointly guarantee differential privacy (DP) and visual privacy—such as face, gender, and skin-tone obfuscation. This paper introduces Video-DPRP, the first sample-level DP-compliant random projection framework for videos, enabling DP-guaranteed video reconstruction without model training. It unifies the evaluation of DP parameters (ε, δ) and quantifiable visual privacy metrics. Leveraging SVD to extract right singular vectors, Video-DPRP integrates noise injection and random projection to preserve utility while enforcing privacy. On UCF101 and HMDB51, it incurs <3.2% accuracy degradation in action recognition. Compared to baselines, it improves facial, gender, and skin-tone privacy protection by up to 41.7%, significantly outperforming state-of-the-art DP and visual privacy methods.

Technology Category

• Computer Vision: Bias, Fairness & Privacy
• Machine Learning: Privacy
• Natural Language Processing: Ethics — Bias, Fairness, Transparency & Privacy

Application Category

• Security and Privacy: Data transparency and provenance
• User Modeling, Personalization and Recommendation: User privacy protection in personalized systems
• Responsible Web: Data and user privacy-enhancing technologies for the Web

Abstract

Considerable effort has been made in privacy-preserving video human activity recognition (HAR). Two primary approaches to ensure privacy preservation in Video HAR are differential privacy (DP) and visual privacy. Techniques enforcing DP during training provide strong theoretical privacy guarantees but offer limited capabilities for visual privacy assessment. Conversely methods, such as low-resolution transformations, data obfuscation and adversarial networks, emphasize visual privacy but lack clear theoretical privacy assurances. In this work, we focus on two main objectives: (1) leveraging DP properties to develop a model-free approach for visual privacy in videos and (2) evaluating our proposed technique using both differential privacy and visual privacy assessments on HAR tasks. To achieve goal (1), we introduce Video-DPRP: a Video-sample-wise Differentially Private Random Projection framework for privacy-preserved video reconstruction for HAR. By using random projections, noise matrices and right singular vectors derived from the singular value decomposition of videos, Video-DPRP reconstructs DP videos using privacy parameters ($epsilon,delta$) while enabling visual privacy assessment. For goal (2), using UCF101 and HMDB51 datasets, we compare Video-DPRP's performance on activity recognition with traditional DP methods, and state-of-the-art (SOTA) visual privacy-preserving techniques. Additionally, we assess its effectiveness in preserving privacy-related attributes such as facial features, gender, and skin color, using the PA-HMDB and VISPR datasets. Video-DPRP combines privacy-preservation from both a DP and visual privacy perspective unlike SOTA methods that typically address only one of these aspects.

Problem

• Develops a model-free approach for visual privacy in videos using differential privacy.
• Evaluates privacy-preserving techniques for human activity recognition using differential and visual privacy assessments.
• Combines differential privacy and visual privacy to preserve attributes like facial features and gender in videos.

Innovation

• Video-DPRP: Differentially Private Random Projection framework
• Combines differential privacy with visual privacy assessment
• Reconstructs videos using noise matrices and SVD