Video analytics use cases to manage COVID-19 response

Four months into 2020, the world is facing a grave global health crisis: the outbreak of a novel coronavirus respiratory disease, COVID-19. Can new digital technology be used to help identify and mitigate the impact of COVID-19? Thankfully, the answer is yes.

One technology now in the forefront of the global emergency is the use of real-time video analytics. With typical use cases that include retail fraud prevention and crowd management, facial recognition, and object detection and classification, the video analytics market is expected to grow from US$2.5 billion in 2018 to 12 billion by 2026, for a compound annual growth rate of 23%. Video analytics is now being applied to innovative use cases, like fever detection, social distancing, and contact tracing. These use cases are expected to further drive the market for video analytics solutions.

Fever detection systems can perform large-area detection in an epidemic and quickly identify those with elevated body temperature, with an accuracy of +/- 1 degree, to help mitigate its impact. Some systems use deep learning models to zoom in on a subject’s inner eye, which is most reflective of the body’s temperature. Infrared and visible video images are then transmitted to AI-based analytics platforms for analysis in real-time. This large-scale body temperature screening can be performed in airports, rail stations, schools, hotels, retail properties, grocery stores, and hospitals.

Social distancing systems combine video footage from existing street cameras with AI vision models to track whether distancing protocols are being followed, assigning each location a physical distancing index (PDI) score. PDI scores can help authorities understand how people are responding to the coronavirus and public health recommendations and orders.

Contact tracing is a core disease control measure for early identification of people who have been potentially exposed. The challenge is that it is labor-intensive, requiring huge teams and significant time. Using real-time video streams, AI-based systems can identify people who have sneezed or coughed, and facilitate contacting people who may have come in contact with them.

Video analytics solutions using AI and deep learning can be deployed on-premise or in the cloud, and from the edge to the core, based on use case. Megh Computing offers a real-time analytics platform to support these new use cases using our ultrahigh-performance Deep Learning Engine (DLE) powered by FPGAs, the fastest and most advanced inferencing engine in the world.



Share this page