Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

PPEDCRF: Privacy-Preserving Enhanced Dynamic CRF for Location-Privacy Protection for Sequence Videos with Minimal Detection Degradation

Bo Ma, Jinsong Wu, Weiqi Yan, Catherine Shi, Minh Nguyen

TL;DR

PPEDCRF is proposed, a privacy-preserving enhanced dynamic conditional random field framework that injects calibrated perturbations only into inferred location-sensitive background regions while preserving foreground detection utility.

Abstract

Dashcam videos collected by autonomous or assisted-driving systems are increasingly shared for safety auditing and model improvement. Even when explicit GPS metadata are removed, an attacker can still infer the recording location by matching background visual cues (e.g., buildings and road layouts) against large-scale street-view imagery. This paper studies location-privacy leakage under a background-based retrieval attacker, and proposes PPEDCRF, a privacy-preserving enhanced dynamic conditional random field framework that injects calibrated perturbations only into inferred location-sensitive background regions while preserving foreground detection utility. PPEDCRF consists of three components: (i) a dynamic CRF that enforces temporal consistency to discover and track location sensitive regions across frames, (ii) a normalized control penalty (NCP) that allocates perturbation strength according to a hierarchical sensitivity model, and (iii) a utility-preserving noise injection module that minimizes interference to object detection and segmentation. Experiments on public driving datasets demonstrate that PPEDCRF significantly reduces location-retrieval attack success (e.g., Top-k retrieval accuracy) while maintaining competitive detection performance (e.g., mAP and segmentation metrics) compared with common baselines such as global noise, white-noise masking, and feature-based anonymization. The source code is in https://github.com/mabo1215/PPEDCRF.git

PPEDCRF: Privacy-Preserving Enhanced Dynamic CRF for Location-Privacy Protection for Sequence Videos with Minimal Detection Degradation

TL;DR

PPEDCRF is proposed, a privacy-preserving enhanced dynamic conditional random field framework that injects calibrated perturbations only into inferred location-sensitive background regions while preserving foreground detection utility.

Abstract

Dashcam videos collected by autonomous or assisted-driving systems are increasingly shared for safety auditing and model improvement. Even when explicit GPS metadata are removed, an attacker can still infer the recording location by matching background visual cues (e.g., buildings and road layouts) against large-scale street-view imagery. This paper studies location-privacy leakage under a background-based retrieval attacker, and proposes PPEDCRF, a privacy-preserving enhanced dynamic conditional random field framework that injects calibrated perturbations only into inferred location-sensitive background regions while preserving foreground detection utility. PPEDCRF consists of three components: (i) a dynamic CRF that enforces temporal consistency to discover and track location sensitive regions across frames, (ii) a normalized control penalty (NCP) that allocates perturbation strength according to a hierarchical sensitivity model, and (iii) a utility-preserving noise injection module that minimizes interference to object detection and segmentation. Experiments on public driving datasets demonstrate that PPEDCRF significantly reduces location-retrieval attack success (e.g., Top-k retrieval accuracy) while maintaining competitive detection performance (e.g., mAP and segmentation metrics) compared with common baselines such as global noise, white-noise masking, and feature-based anonymization. The source code is in https://github.com/mabo1215/PPEDCRF.git
Paper Structure (18 sections, 3 theorems, 20 equations, 7 figures, 1 table, 1 algorithm)

This paper contains 18 sections, 3 theorems, 20 equations, 7 figures, 1 table, 1 algorithm.

Table of Contents

  1. Introduction
  2. Threat Model
  3. Research Range
  4. Related Work
  5. Motivations and Contributions
  6. Our Proposed Methods
  7. Overview Architecture
  8. Subsampling for Privacy Noise Data
  9. Observation and classification from sequence images with conditional random fields
  10. Restructure the discernibility matrix (DM) with Normalized Control Penalty(NCP) method
  11. Feature-space Noise Injection and Reconstruction
  12. Noise generating
  13. Privacy-preserving Enhanced Dynamic Conditional Random Field(PPEDCRF) Algorithm
  14. Experimental Evaluations
  15. The performance of PPEDCRF and NCP for object detection
  16. ...and 3 more sections

Key Result

Lemma 1

In transfer function the cumulative sum of predictive region $\omega$ as $\Delta \omega_{t-1}$ and sensitive region $\rho_{1:T}$ from states 1 to $T$ according to their privacy score, those results can be deduced from the image data sets and hyper parameter $P_n (\eta,\tau,G)$.

Figures (7)

  • Figure 1: Images taken by a dashcam can be matched with an image from Google Street View.
  • Figure 2: The Architecture of Privacy-preserving Enhanced Dynamic Conditional Random Field(PPEDCRF) for Segmentation and Object Detection
  • Figure 3: The performance of objects detection with and without PPEDCRF+NCP
  • Figure 4: Predict Result on Privacy-preserving Enhanced Dynamic Conditional Random Field (PPEDCRF)
  • Figure 5: Comparison of privacy-preserving performance for detection task
  • ...and 2 more figures

Theorems & Definitions (6)

  • Definition 1: Privacy Cost(Budget)
  • Lemma 1: Transfer function for sensitive region and privacy score
  • Lemma 2: Masked Gaussian noise increases sensitive-feature distance
  • proof
  • Theorem 1: Expected similarity reduction under masked feature noise
  • proof