How behavior analytics can be useful during covid-19

Needless to say, the covid-19 pandemic has taken a huge toll across the globe, forcing end user entities all over the world to issue lockdown, shelter-in-place and social distancing orders to contain the disease. Behavior analytics that can help with the enforcement of these orders, then, can be useful.

The coronavirus has become a global pandemic, infecting millions and killing tens of thousands. This has prompted end user organizations to put in place certain measures to effectively contain the disease and prevent it from spreading further. Social distancing rules for example are issued across the globe. Here in Taiwan, masks are to be worn at all times on rail transportation.

One solution that can help effectively enforce these disease control rules is behavior analytics, which take video feeds from cameras and analyze whether certain behaviors are exhibited by human objects in the video. For end user entities that need to check if people are at certain places they are not supposed to be or that need to prepare for possible acts of looting and vandalism during business shutdowns and economic hardships, behavior analytics can come in handy.

"Behavior recognition analytics can be used to detect proximity in groups, crowds and gatherings (different group sizes), movement of people and vehicles where such movement is restricted, identifying whether people are wearing masks and hazmat suits, standing or waiting in lines while keeping distance from one another, as well as helping people in distress for example when falling or lying on the ground,” Maya Scheyer, VP of Business Development and Sales at viisights.
 

Monitoring different behaviors

Based in Israel, viisights offers different types of AI-based, server-side analytics with different applications. According to Scheyer, in the case of covid-19 control and prevention, viisights is targeting municipal administration, medical and healthcare centers, law enforcement, enterprises and campuses to push their behavior analytics which are capable of the following:

• Crowd and group behavior recognition: Assisting in locating and enforcing crowds gathering in violation of regulations and alerting on specific behaviors such as crowd gathering and crowd running;
• Suspicious activity recognition: Persons running or persons in distress such as falling or lying on the ground which may happen in elderly facilities or medical institutions can be detected;
• Violence and violent activity recognition: With the prospect of looting, vandalism and robberies on the horizon, viisights is able to recognize violent and suspicious activities such as people fighting, stabbing and carrying weapons;
• Perimeter protection and safety: Detecting movement of people and vehicles in unauthorized areas and timeframes (for example, some localities have a curfew);
• Data for statistics and heatmap: The company offers data and connectivity for predictive analytics third-party party products, enabling end users to build predictive models for prediction of possible outbreak areas.

According to Scheyer, viisights’ technology stands out in that it is uses video clips rather than discrete images for training its core AI engine. “To date, most of video analytics systems still base their product’s features on static analysis of objects from images using image recognition. Products which are built on using such object classification technology are extremely limited. For example, they will not be able to recognize behavioral events in a video, such as people fighting or car collisions, because such behaviors can’t accurately be deduced from analyzing a single static image/frame in large scale,” she said. “viisights is based on behavioral understanding of the video content. We can extract more meaningful data from the huge amount of captured video content and transform it to actionable insights, eventually justifying the massive investment in video surveillance infrastructure.”