Charlotte Dryden, Director of the Visual Computing Developer Solutions team at Intel, presents the “Balancing Safety, Convenience and Privacy in the Era of Ubiquitous Cameras” tutorial at the May 2018 Embedded Vision Summit.
Computer vision-enabled cameras are proliferating rapidly and will soon be ubiquitous – in, on and around vehicles, homes, toys, stores, public transit, schools, restaurants and more. Clearly, this offers tremendous benefits in terms of safety, security, convenience and efficiency. But what about privacy? Are we doomed to give up our privacy as cameras proliferate? Possibly, but not necessarily. Many of the same technologies that are fueling the proliferation of visual intelligence can also be used to enhance privacy, if product developers choose to do so, and if consumers, enterprises and governments prioritize privacy.
For example, accelerating innovation in sensors means that system designers have many choices of sensor types beyond the typical CMOS image sensor, enabling engineers to choose sensor types that capture only the information that is required for the application. And rapid progress in embedded processors – combined with efficient, accurate algorithms – makes it increasingly feasible to consume images at the edge or in the fog, and then discard them, retaining only the required meta-data. In this talk, Dryden explores trade-offs related to privacy in a world filled with connected cameras.