Using Machine Learning to Fight COVID-19

Haozhi Zhang took the drive to encourage people to wear masks and practice social distancing to battle COVID-19 as a personal challenge. 

Zhang, a student in Northwestern Engineering's Master of Science in Robotics (MSR) program, applied the lessons he’s learned in the program to create a visual tool that can determine if people are complying with Centers for Disease Control and Prevention (CDC) safety guidelines. 

“It is important to remind society to keep social distance and wear masks, so that was the initial motivation,” Zhang said. “I am personally interested in applying machine learning to solve real-world problems. This topic combined my motivation and interests.” 

The tool was created for Zhang's independent project, one of the requirements to graduate from MSR. Zhang used machine learning algorithms and a RealSense depth camera to determine if people were wearing masks and maintaining safe distances from others. By the time it was finished, the tool could reliably detect faces in real-time live streams and determine whether they were masked and how close together they were. 

“The most exciting part was when the combination of several technologies worked well in live streaming,” Zhang said. “The moment when I used them to detect human faces and classify whether they were wearing masks was exciting to me because I could see the direct result in reality, not just as a conceptual idea.” 

Getting to that point wasn’t easy, Zhang said. The first step was the most difficult. 

“The biggest challenge was to get started with the whole pipeline and become familiar with machine learning algorithms and libraries,” he said. “The beginning of a project is always a struggle – but it's also essential.” 

The project is independent by nature, but Zhang was not alone in his work. He frequently turned to his classmates and MSR faculty for guidance. That collaborative component is a hallmark of MSR and its cohort model.

“Those discussions not only give you valuable suggestions," Zhang said, "but they also help clear your own mind.” 

Zhang was encouraged by what he accomplished during the class. 

“I was impressed by the progress I made during one quarter,” he said. “The result of the project met my expectations.” 

Meanwhile, the MSR program is exceeding his expectations. Zhang said the work on this project put him in a great place to finish the program strong and use what he’s learned to make a difference in the robotics industry when he graduates. 

“The project definitely built my confidence for the final project during the last quarter in MSR," he said. "It shaped my knowledge and coding skills to apply machine learning algorithms to solve real-world problems."