AI Solution for COVID
Combating Covid-19 at KAUST with AI-powered computer vision platform VIZARD
Share on facebook
Share
Share on twitter
Retweet
Share on linkedin
Share
Share on whatsapp
Send
The Scenario
Earlier this year, the KAUST Smart team was looking for technology solutions to control the spread of COVID-19. So we, the UnitX team, came up with an AI solution called VIZARD. It is an artificial intelligence based platform that uses proprietary Computer Vision software to provide real-time insights to the user from videos; video sources could be a CCTV camera or any external camera. The Computer Vision software analyzes videos and can be trained to identify objects such as people, masks, cars, goggles, registration plates, helmets, and so much more, entirely customizable as per the client’s requirements. Vizard is not only built for safety and hazard monitoring but has a plethora of use cases such as – crowd and queue management, foreign object detection, and 24/7 site inspection, etc. For KAUST’s purposes, the Vizard platform currently measures compliance with wearing masks at their Library.The Solution
Vizard was first deployed at the KAUST Campus Library, the world’s largest LEED Platinum project, which was awarded the ‘Top 10 Green Projects’ by the American Institute of Architects in 2011. After being closed for 3 months due to COVID-19, the management wanted to open it again while ensuring compliance with safety regulations. UnitX worked with KAUST Smart and KAUST HSE to build a customized solution. This Vizard pilot delivers real-time alerts while adhering to the data privacy and information security guidelines of KAUST. The metrics monitored by Vizard, such as the compliance metrics and alerts, have been set for KAUST’s requirements. The current deployment consists of a camera at the library’s entrance that collects visual feed on people entering the library. Our AI software then identifies whether individuals are wearing masks and complying with the recommended space for social distancing, 1 meter between individuals. If someone is not adhering to the safety regulations, Vizard flags an alert in real-time on the dashboard that the KASUT executives could view and act on.The Technology & Challenges Faced
Vizard’s artificial intelligence software consists of deep learning models for computer vision, with “model” referring to the software and algorithms used. For the deep learning model to learn objects and identify them, it relies on a neural network made up of “neurons.” Neurons are interconnected layers of software-based calculations that can ingest large amounts of image or video data, process them through the many layers of the network, and learn how to identify objects or features such as faces or masks. Once the deep learning model knows how to identify objects or features correctly, it can detect them again from a new image or video data. The initial model we deployed detected the distances between individuals. Anything less than the recommended distance of 1 meter is immediately flagged in real-time, sending out alerts on the platform. The project’s challenge was not model development; we developed it quickly and efficiently. Instead, the challenge was to make sure the system is protected. Therefore we took extra care to ensure high-end data security and data privacy. Today, the device and software are secure and are all done on-premise using KAUST Wifi and storage servers. Based on KAUST Smart’s requirements, we deployed another model that learned mask detection a few days after the initial pilot project launch. That model is successfully running for 2 months now and updating the HSE Vizard Dashboard in real-time.What’s Next
In collaboration with KAUST Smart, not only did we have the opportunity to contribute to the community but also further develop its technology. The Vizard pilot project deployment has been successful, as well as stable and secure. Using what we have learned from this pilot, we are ready to build more complex AI capabilities and advanced features for organizations across industries such as Manufacturing, Retail, Logistics & transportation, Hospitality etc.Follow us to stay tuned
Linkedin | Twitter | YouTube
Share on facebook
Share
Share on twitter
Retweet
Share on linkedin
Share
Share on whatsapp
Send
