The surveillance support system to ensure a safe distance from the Covid-19 epidemic from the application of artificial intelligence (AI) research of students has been successfully tested.
On the morning of April 26, Dr. Nguyen Tan Tran Minh Khang - Vice Rector of the University of Information Technology, Vietnam National University Ho Chi Minh City - said a group of students and researchers specializing in computer vision in the laboratory. Multimedia communication (MMLab) of the school has built a monitoring support system to ensure a safe distance through the camera.
Experimental images of the system on actual data from CCTV - (Photo: The research team provided).
"The team consists of students and researchers who have applied advanced techniques in the field of computer vision and machine learning (Deep Learning) to develop this test system. support for anti-epidemic of Covid-19 " , Mr. Khang said.
The research team consists of members: Le Quang Hung (third year student, Faculty of Computer Science), researcher Nguyen Ngoc Thua (MMLab), under the guidance of MA. Do Van Tien.
According to the research team, during the Covid-19 pandemic, people were recommended by the Ministry of Health to limit contact, go to crowded places and keep a distance of at least 2m from each other when communicating. Many places have applied different solutions such as printing, drawing location markers, using marking tools, or with the guidance of volunteers.
However, the above solution is difficult to expand or lack flexibility when implemented under different conditions. Most importantly, surveillance ensures a safe distance mainly based on people's observations, or people's self-awareness.
Support for social spacing through CCTV introduced by the research team.
"Prior to that fact, we immediately thought of using AI, exploiting data from CCTV that is now quite popular in public places such as companies, schools, squares, hospitals . to build a distance monitoring support system " , Le Quang Hung shared.
With the input being the image obtained from the CCTV, the system will automatically locate and estimate the distance between people in the frame. From the computed information, the system will output warnings when the distance between people is not guaranteed.
According to MA Do Van Tien, the system can support and reduce the supervision work of volunteers, especially in public places with large spaces such as schools, factories, and hospitals.
"With the current version, the system is capable of executing at near-real-time speed. The warnings given in a frame are classified into two categories: risk and high risk based on contact distance. between objects, " said Tien.
The system will need to continue to be developed and refined to improve accuracy under various application conditions.
"The research team is ready to share and collaborate with interested individuals and units to develop, integrate or test deploy this application," Mr. Thua said.