Detecting and Mitigating Adversarial Attacks on Covid-19 Monitoring Solutions

COVID-19 outbreak has changed the world and forced majority of the population to lockdown and isolation. With the lockdowns easing, and things returning to the new normal, it is important to follow the isolation protocols such as social distancing and wearing of masks for containing the outbreak. While there is no denying the efficacy of the protocols, there has been some backlash in the public for these policies including a large number of protests against the COVID-19 protocols in the US [1] [2]. Despite privacy concerns and protests these protocols are an important tool for public safety as they ensure that the number of active infections remain manageable and health care systems do not collapse. As Canadians get back to work there is a need to efficiently monitor these protocols for public safety. Machine vision-based technologies are helpful, and a growing number of companies are offering such monitoring solutions, which ensure that certain conditions are met (e.g., the visitor is wearing a face mask). Because of the negative sentiment in some portion of the population there is a possibility of adversaries who would try to circumvent these monitoring solutions.

Faculty Supervisor:

Hassan Khan

Student:

Sohail Habib

Partner:

Gradient Ascent Inc.

Discipline:

Computer science

Sector:

Professional, scientific and technical services

University:

University of Guelph