BLINDSPOT: Enabling Bystander-Controlled Privacy Signaling for Camera-Enabled Devices

2025-12-12

AI Summary

Camera-equipped devices (e.g., smartphones, AR glasses) pose real-time privacy threats to bystanders’ faces and imagery. Method: This paper proposes a registration-free, locally executed privacy preference signaling system that avoids leakage of sensitive information. It introduces the first tri-modal active signaling framework—comprising on-device gesture recognition, enhanced visible light communication (VLC), and a novel ultra-wideband (UWB) protocol—augmented by a geometric consistency verification mechanism to prevent identity spoofing. The entire system is implemented on commodity smartphones. Results: Empirical evaluation demonstrates feasible accuracy across diverse distances, dynamic lighting conditions, and motion scenarios, with sub-millisecond end-to-end latency. The core contribution is the first lightweight, privacy-preserving, and forgery-resistant real-time framework enabling bystanders to autonomously control their visual privacy.

Technology Category

• Computer Vision: Bias, Fairness & Privacy
• Machine Learning: Privacy
• Humans and AI: Interaction Techniques and Devices

Application Category

• Security and Privacy: Large-scale security measurements
• User Modeling, Personalization and Recommendation: User privacy protection in personalized systems
• Systems and Infrastructure for Web, Mobile and WoT: Energy management for devices in mobile Web and WoT environments

Abstract

Camera-equipped mobile devices, such as phones, smart glasses, and AR headsets, pose a privacy challenge for bystanders, who currently lack effective real-time mechanisms to control the capture of their picture, video, including their face. We present BlindSpot, an on-device system that enables bystanders to manage their own privacy by signaling their privacy preferences in real-time without previously sharing any sensitive information. Our main contribution is the design and comparative evaluation of three distinct signaling modalities: a hand gesture mechanism, a significantly improved visible light communication (VLC) protocol, and a novel ultra-wideband (UWB) communication protocol. For all these modalities, we also design a validation mechanism that uses geometric consistency checks to verify the origin of a signal relative to the sending bystander, and defend against impersonation attacks. We implement the complete system (BlindSpot) on a commodity smartphone and conduct a comprehensive evaluation of each modality's accuracy and latency across various distances, lighting conditions, and user movements. Our results demonstrate the feasibility of these novel bystander signaling techniques and their trade-offs in terms of system performance and convenience.

Problem

• Enables bystanders to control privacy in real-time for camera devices.
• Develops and compares three signaling methods for privacy preferences.
• Implements validation to prevent impersonation and ensure signal authenticity.

Innovation

• Hand gesture mechanism for privacy signaling
• Improved visible light communication protocol
• Novel ultra-wideband communication protocol with validation