Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

TSDCRF: Balancing Privacy and Multi-Object Tracking via Time-Series CRF and Normalized Control Penalty

Bo Ma, Jinsong Wu, Weiqi Yan

Abstract

Multi-object tracking in video often requires appearance or location cues that can reveal sensitive identity information, while adding privacy-preserving noise typically disrupts cross-frame association and causes ID switches or target loss. We propose TSDCRF, a plug-in refinement framework that balances privacy and tracking by combining three components: (i) $(\varepsilon,δ)$-differential privacy via calibrated Gaussian noise on sensitive regions under a configurable privacy budget; (ii) a Normalized Control Penalty (NCP) that down-weights unstable or conflicting class predictions before noise injection to stabilize association; and (iii) a time-series dynamic conditional random field (DCRF) that enforces temporal consistency and corrects trajectory deviation after noise, mitigating ID switches and resilience to trajectory hijacking. The pipeline is agnostic to the choice of detector and tracker (e.g., YOLOv4 and DeepSORT). We evaluate on MOT16, MOT17, Cityscapes, and KITTI. Results show that TSDCRF achieves a better privacy--utility trade-off than white noise and prior methods (NTPD, PPDTSA): lower KL-divergence shift, lower tracking RMSE, and improved robustness under trajectory hijacking while preserving privacy. Source code in https://github.com/mabo1215/TSDCRF.git

TSDCRF: Balancing Privacy and Multi-Object Tracking via Time-Series CRF and Normalized Control Penalty

Abstract

Multi-object tracking in video often requires appearance or location cues that can reveal sensitive identity information, while adding privacy-preserving noise typically disrupts cross-frame association and causes ID switches or target loss. We propose TSDCRF, a plug-in refinement framework that balances privacy and tracking by combining three components: (i) -differential privacy via calibrated Gaussian noise on sensitive regions under a configurable privacy budget; (ii) a Normalized Control Penalty (NCP) that down-weights unstable or conflicting class predictions before noise injection to stabilize association; and (iii) a time-series dynamic conditional random field (DCRF) that enforces temporal consistency and corrects trajectory deviation after noise, mitigating ID switches and resilience to trajectory hijacking. The pipeline is agnostic to the choice of detector and tracker (e.g., YOLOv4 and DeepSORT). We evaluate on MOT16, MOT17, Cityscapes, and KITTI. Results show that TSDCRF achieves a better privacy--utility trade-off than white noise and prior methods (NTPD, PPDTSA): lower KL-divergence shift, lower tracking RMSE, and improved robustness under trajectory hijacking while preserving privacy. Source code in https://github.com/mabo1215/TSDCRF.git
Paper Structure (31 sections, 17 equations, 13 figures, 3 tables)

This paper contains 31 sections, 17 equations, 13 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Motivations.
  3. Threat model
  4. Attack 1: Feature-stealing / identity spoofing
  5. Attack 2: Track and hijack the target orbit
  6. Limitations of Differential Privacy mechanism for object tracking
  7. Contributions
  8. Paper organization.
  9. Related Works
  10. Privacy-preserving tracking and detection
  11. Object detection for tracking
  12. Privacy-preserving in object detection
  13. Object tracking and self-attention
  14. Tracking object lost and occlusion
  15. Trajectory integrity and consistency Check
  16. ...and 16 more sections

Figures (13)

  • Figure 1: Segmentation and tracking results. Top: three consecutive frames from MOT16. Bottom: segmentation and trajectory after target identification.
  • Figure 2: Tracking Result For Vehicle
  • Figure 3: Flowchart of the proposed framework: detector $\rightarrow$ tracker $\rightarrow$ privacy module $\rightarrow$ TSDCRF refinement.
  • Figure 4: Motion trajectory prediction with the CRF (Fig. 4). The time-series CRF enforces temporal consistency and reduces position deviation after privacy noise.
  • Figure 5: The performance of the proposed methods. (a) Defense under trajectory hijacking attack; (b) defense under blind target object detection attack.
  • ...and 8 more figures