Learning Privacy-preserving Optics for Human Pose Estimation

We design the camera lens to perform human pose estimation while preserving users' privacy.

Carlos Hinojosa

Universidad Industrial de Santander

Juan Carlos Niebles

Stanford University

Henry Arguello

Universidad Industrial de Santander

Abstract

Cameras are everywhere! How to develop privacy-preserving vision systems?

Cameras are everywhere! How to develop privacy-preserving vision systems?

The widespread use of always-connected digital cameras in our everyday life has led to increasing concerns about the users' privacy and security. How to develop privacy-preserving computer vision systems? In particular, we want to prevent the camera from obtaining detailed visual data that may contain private information. However, we also want the camera to capture useful information to perform computer vision tasks. Inspired by the trend of jointly designing optics and algorithms, we tackle the problem of privacy-preserving human pose estimation by optimizing an optical encoder ( hardware-level protection) with a software decoder (convolutional neural network) in an end-to-end framework. We introduce a visual privacy protection layer in our optical encoder which, parametrized appropriately, enables the optimization of the point spread function (PSF) of the camera lens. We validate our approach with extensive simulations and a prototype camera. We show that our privacy-preserving deep optics approach successfully degrades or inhibits private attributes while maintaining important features to perform human pose estimation.

Method Overview

Results

Qualitative results on example COCO images. We compare our proposed privacy-preserving pose estimation results using the optimized lens with the Non-privacy approach using a standard lens. The last two columns depict failure cases where we fail to estimate the pose of far distant people.

Qualitative results on example COCO images. We compare our proposed privacy-preserving pose estimation results using the optimized lens with the Non-privacy approach using a standard lens. The last two columns depict failure cases where we fail to estimate the pose of far distant people.

Lab Implementation

ICCV 2021 (Oral) Video

Social Media Posts

Cite this work

@InProceedings{Hinojosa_2021_ICCV,
    author    = {Hinojosa, Carlos and Niebles, Juan Carlos and Arguello, Henry},
    title     = {Learning Privacy-Preserving Optics for Human Pose Estimation},
    booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)},
    month     = {October},
    year      = {2021},
    pages     = {2573-2582}
}