Towards long-term large-scale visual health monitoring using Cyber Glasses

More than 30% of the world population have vision defects, for some of which causes are still unclear. Visual health monitoring for detection, prevention, and treatment is possible but still very limited due to limited access to expensive specialized equipment and domain experts. Therefore, it is difficult to provide long-term visual health monitoring for a large population. In this paper, we present the design and evaluation of Cyber Glasses, low-cost computational glasses as a step toward longterm large-scale human visual health monitoring. At the core of Cyber Glasses are three key novel contributions: an integration of low-cost commercial off the shelf (COTS) components, an adaptive data collection mechanism taking into account tradeoff's between sensing accuracy, latency, memory, and energy, and a suit of energy efficient algorithms to reduce sensor data size and to extract meaningful human vision information to highlevel applications. We conduct a number of experiments to verify the feasibility of Cyber Glasses to enable long-term large-scale human visual health monitoring.