Work

Re-imagining Wearable Visual Systems for Context-Aware Computing

Public

Downloadable Content

Download PDF

Wearable visual systems, such as ego-centric wearable cameras, have failed to integrate into everyday life. We have witnessed the abandonment of wearable visual systems as consumer devices (e.g., Google Glass) and as research tools (e.g., SenseCam). While it is natural for some technologies to die out, visual wearable systems are indispensable tools that advance the entire field of context-aware wearables. Particularly, visual wearable systems can push the domain forward by (1) enabling interactive application that needs access to visual data and by (2) acting as a sensemaking tool that helps in contextualizing data collected with other wearables that use non-visual sensors (e.g., accelerometers), allowing for further granularity in data set creation and improving the accuracy of data annotations used in machine learning-based model training and validation. Both technical and societal challenges lead to the abandonment of wearable cameras. First, there are substantive privacy concerns and stigma around wearing cameras. Even when privacy measures are in place, many people refuse to wear them because they are regarded as invasive tools that cause discomfort, especially to bystanders. Second, it is hard to process images captured by wearable cameras. Performing typical image capturing and processing tasks requires significant computational and energy resources, making it impractical to develop potential methods that can truly enhance the privacy of these devices. My work addresses these challenges by (1) providing a framework that helps in understanding and manipulating design factors to reduce privacy and stigma concerns, (2) developing a practical obfuscation method to enhance privacy in visual data collection while maintaining utility, (3) building novel and efficient dual-imaging visual systems enables practical data collection and processing, (4) investigating and building alternatives to light-based visual systems that can provide rich enough information for many human-centered applications without sacrificing privacy, efficiency or energy. By centering visual systems and addressing the ongoing tensions, my work paves the way for a sustainable future for both visual and non-visual wearable systems.

Creator
DOI
Subject
Language
Alternate Identifier
Date created
Resource type
Rights statement

Relationships

Items