What If Blind People Wear a Camera That Can Detect You in Public Spaces?
Understanding social acceptance of assistive wearable cameras from two different perspectives.
Wearable cameras can serve as assistive technology for blind people to detect pedestrians as it can provide access to their visual information, such as their approximate distance, relative position, gender, and facial expressions. Public use of such technology, however, may raise privacy concerns for bystanders due to the always-on nature of this technology.
To understand design factors for assistive wearable cameras that are more acceptable to not only blind people but also sighted bystanders, we explore public notions of privacy and societal consent around such assistive wearable cameras. In particular, we show participants videos of such scenarios and further situate them to experience such scenarios personally. One way of refining our understanding of the world is through experience. We may have been told that “ice is cold,” and even understood it, but we really get it once we touch ice. Experience can provide a better understanding of learned concepts. We want to explore how people’s attitudes toward assistive technologies are affected through experience.
Study on social acceptance of wearable cameras for pedestrian detection
Our research team looks at the social acceptance of wearable cameras for blind pedestrians by conducting two different studies: an online survey and an in-person experiment. In the online survey, 206 sighted people, recruited via Amazon Mechanical Turk, watched videos showing a blind and a sighted person pass each other in a corridor. Videos were shown from two perspectives: the passerby perspective and the blind user perspective. The former shows a blind user wearing the assistive technology, and the latter shows the pedestrian captured by the camera and information detected by an algorithm.
In the in-person experiment, we recruited 10 blind participants and 40 sighted participants (four sighted participants per blind participant). The scenario is similar to that shown in the videos of the online survey. Blind participants ask a pedestrian for the nearest office number while wearing a pair of smart glasses for detecting a pedestrian and a Bluetooth earbud for receiving audible information, such as distance, position, head pose, approximate age, and gender of the nearby pedestrian. On the other hand, the sighted participants simply walk in the same corridor. After walking and interacting with blind users, sighted participants complete the same questionnaires as in the online survey.
Virtual learning vs. personal experience
While sighted participants in the online survey were exposed to this assistive technology via the videos, sighted participants in the in-person study directly saw a blind user wearing the smart glasses and learned about how the technology works later on. Comparing responses from these two different groups, we see that there is a difference in their acceptance toward the technology. Although both tend to be rather positive towards the assistive use of wearable cameras, in-person sighted participants feel a bit more uncomfortable about it than online sighted participants.
Blind people vs. sighted people
What are the opinions of blind people and sighted people on this technology? Do these two parties agree on the public use of this technology without any conflicts of interest? Our work highlights design factors that may influence societal consent. There is agreement on several detected visual characteristics. For example, neither party cared for the age or ethnicity of a person being detected by the assistive technology. One blind participant remarked,
“If there is a baby one year old, (the baby) can’t tell you what the number on a door is. Other than that, (age estimation) doesn’t make any difference.”
Their disagreement is on visual attributes related to head pose (eye gaze) and facial expressions. Blind participants prefer to learn about a nearby pedestrian’s head pose to know whether a passerby is looking at them. They also value the estimation of facial expressions that may serve as indicators for social interaction. Sighted participants, however, are not keen on this information being detected from them.
Our insights
- Overall, sighted participants show positive attitudes toward wearable cameras for pedestrian detection that benefit blind people. This only extends to technology where the camera is visible and the extracted information is agreeable.
- Future studies that examine social negotiation of technology use between different parties should consider the experiential use cases as they may result in different views.
This article is a summary of our CHI 2020 publication authored by Kyungjun Lee, Daisuke Sato, Saki Asakawa, Hernisa Kacorri, and Chieko Asakawa. All visualizations are created by Tzu-Chai Yeh, a 2nd-year HCIM student at the UMD iSchool. Read our full paper below:
- Kyungjun Lee, Daisuke Sato, Saki Asakawa, Hernisa Kacorri, and Chieko Asakawa. (2020). Pedestrian Detection with Wearable Cameras for the Blind: A Two-way Perspective. In Proceedings of the ACM Conference on Human Factors in Computing Systems Proceedings (CHI 2020), April 25–30, 2020, Honolulu, Hawai’i, USA. ACM, New York, NY, USA.
Also watch our 30-second preview video:
