From Pixels to Privacy: Temporally Consistent Video Anonymization via Token Pruning for Privacy Preserving Action Recognition

• URL: http://arxiv.org/abs/2603.26336v1
• Date: Fri, 27 Mar 2026 11:57:51 GMT
• Title: From Pixels to Privacy: Temporally Consistent Video Anonymization via Token Pruning for Privacy Preserving Action Recognition
• Abstract summary: We propose a video anonymization framework based on systematic video anonymization and privacy features.<n>Our key insight is that attention mechanisms in Vision Transformers can be structured to separate action-relevant information from privacy-sensitive content.<n>Our approach maintains recognition comparable to models trained on raw videos, while substantially reducing privacy leakage.
• Score: 14.897629110496219
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Recent advances in large-scale video models have significantly improved video understanding across domains such as surveillance, healthcare, and entertainment. However, these models also amplify privacy risks by encoding sensitive attributes, including facial identity, race, and gender. While image anonymization has been extensively studied, video anonymization remains relatively underexplored, even though modern video models can leverage spatiotemporal motion patterns as biometric identifiers. To address this challenge, we propose a novel attention-driven spatiotemporal video anonymization framework based on systematic disentanglement of utility and privacy features. Our key insight is that attention mechanisms in Vision Transformers (ViTs) can be explicitly structured to separate action-relevant information from privacy-sensitive content. Building on this insight, we introduce two task-specific classification tokens, an action CLS token and a privacy CLS token, that learn complementary representations within a shared Transformer backbone. We contrast their attention distributions to compute a utility-privacy score for each spatiotemporal tubelet, and keep the top-k tubelets with the highest scores. This selectively prunes tubelets dominated by privacy cues while preserving those most critical for action recognition. Extensive experiments demonstrate that our approach maintains action recognition performance comparable to models trained on raw videos, while substantially reducing privacy leakage. These results indicate that attention-driven spatiotemporal pruning offers an effective and principled solution for privacy-preserving video analytics.

Related papers

• Knowledge Priors for Identity-Disentangled Open-Set Privacy-Preserving Video FER [29.528717718182975]
  We propose a two-stage framework for video-based privacy-preserving FER.<n>We first train an identity-suppression network using intra- and inter-video knowledge priors.<n>A subsequent denoising module restores expression-related information and helps recover FER performance.
  arXiv  Detail & Related papers  (2026-03-22T20:18:31Z)
• Neural Gate: Mitigating Privacy Risks in LVLMs via Neuron-Level Gradient Gating [71.55435880263238]
  We introduce Neural Gate, a novel method for mitigating privacy risks through neuron-level model editing.<n>Our method improves a model's privacy safeguards by increasing its rate of refusal for privacy-related questions.
  arXiv  Detail & Related papers  (2026-03-13T03:03:20Z)
• Privacy Beyond Pixels: Latent Anonymization for Privacy-Preserving Video Understanding [56.369026347458835]
  We introduce a novel formulation of visual privacy preservation for video foundation models that operates entirely in the latent space.<n>Current privacy preservation methods on input-pixel-level anonymization require retraining the entire utility video model.<n>A lightweight Anonym Adapter Module (AAM) removes private information from video features while retaining general task utility.
  arXiv  Detail & Related papers  (2025-11-11T18:56:27Z)
• Activity Recognition on Avatar-Anonymized Datasets with Masked Differential Privacy [64.32494202656801]
  Privacy-preserving computer vision is an important emerging problem in machine learning and artificial intelligence.<n>We present anonymization pipeline that replaces sensitive human subjects in video datasets with synthetic avatars within context.<n>We also proposeMaskDP to protect non-anonymized but privacy sensitive background information.
  arXiv  Detail & Related papers  (2024-10-22T15:22:53Z)
• CausalVE: Face Video Privacy Encryption via Causal Video Prediction [13.577971999457164]
  With the proliferation of video and live-streaming websites, public-face video distribution and interactions pose greater privacy risks. We propose a neural network framework, CausalVE, to address these shortcomings. Our framework has good security in public video dissemination and outperforms state-of-the-art methods from a qualitative, quantitative, and visual point of view.
  arXiv  Detail & Related papers  (2024-09-28T10:34:22Z)
• Diff-Privacy: Diffusion-based Face Privacy Protection [58.1021066224765]
  In this paper, we propose a novel face privacy protection method based on diffusion models, dubbed Diff-Privacy. Specifically, we train our proposed multi-scale image inversion module (MSI) to obtain a set of SDM format conditional embeddings of the original image. Based on the conditional embeddings, we design corresponding embedding scheduling strategies and construct different energy functions during the denoising process to achieve anonymization and visual identity information hiding.
  arXiv  Detail & Related papers  (2023-09-11T09:26:07Z)
• STPrivacy: Spatio-Temporal Tubelet Sparsification and Anonymization for Privacy-preserving Action Recognition [28.002605566359676]
  We present a PPAR paradigm, i.e. spatial, performing privacy preservation from both temporal perspectives, and propose a STPrivacy framework. For first time, our STPrivacy applies vision Transformers to PPAR and regards video as sequence of leakage-temporal tubelets. Because there is no large-scale benchmarks, we annotate five privacy attributes for two of the most popular action recognition datasets.
  arXiv  Detail & Related papers  (2023-01-08T14:07:54Z)
• PrivHAR: Recognizing Human Actions From Privacy-preserving Lens [58.23806385216332]
  We propose an optimizing framework to provide robust visual privacy protection along the human action recognition pipeline. Our framework parameterizes the camera lens to successfully degrade the quality of the videos to inhibit privacy attributes and protect against adversarial attacks.
  arXiv  Detail & Related papers  (2022-06-08T13:43:29Z)
• SPAct: Self-supervised Privacy Preservation for Action Recognition [73.79886509500409]
  Existing approaches for mitigating privacy leakage in action recognition require privacy labels along with the action labels from the video dataset. Recent developments of self-supervised learning (SSL) have unleashed the untapped potential of the unlabeled data. We present a novel training framework which removes privacy information from input video in a self-supervised manner without requiring privacy labels.
  arXiv  Detail & Related papers  (2022-03-29T02:56:40Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.