Security of containerized computer vision applications

Computer vision processing requires many resources that are easily accessible in the Cloud. However, with the growth of the Internet of Things (IoT), sending all the data to the cloud to analyze is undesirable due to delay in response, high bandwidth consumption, reliability issues, and privacy concerns. Edge computing is an attractive solution to address these issues by processing the data at edge nodes. Well-defined nature of IoT applications enable lightweight containers at the edge to run the computer vision applications. However, security of containers running computer vision applications is not studied in detail. In this paper, we investigate security of containerized computer vision application using Docker containers using multiple tools and we design our own containers that provide better security for computer vision applications running on the edge nodes.