Deep Learning-Assisted Visual Sensing to Detect Overcrowding in COVID-19 Infected Cities


The proposed method uses drones to capture information on overcrowding and analyze it in real-time

Scientists from Korea have proposed a deep learning assisted model to detect overcrowding and crowd behavior in COVID-19-affected cities. This model can help slow the spread of infectious diseases by detecting crowd changes in real-time using unmanned aerial vehicles (UAVs) and social monitoring systems (SMS).

Infectious diseases, such as COVID-19, spread easily in crowded areas. Smart visual sensing that can identify overcrowding is, thus, a necessary tool to curb the spread of these infectious diseases. Recently, a team of scientists from Iran, India, and Korea, has proposed a system that uses unmanned aerial vehicles (UAVs), such as drones, and deep learning algorithms to detect overcrowding in cities with approximately 96% accuracy.

Crowded places tend to be a hub for infectious disease transmission. The COVID-19 pandemic has shown us that it is necessary to find ways to manage crowded areas to help curtail the spread of infectious diseases. Unmanned aerial vehicles (UAVs), such as drones, can detect and record environmental conditions at different heights above the ground in real-time. This makes them ideal for detecting overcrowding and abnormal crowd behaviors, such as riots.

To this end, a group of scientists led by Professor Gwanggil Jeon from Incheon National University, Korea, have developed a real-time visual sensing system using deep learning algorithms. “Information and communications (ICT) technologies can help in the management of. In this paper, we propose a real-time system to detect overcrowding and abnormal crowd behavior. The monitoring system detects over-density using UAVs that communicate with a social monitoring system (SMS),” says Prof. Jeon. Their findings were made available online on May 12, 2022, in IEEE Transactions on Industrial Informatics and published in Volume 19, Issue 1 of the journal on January 1, 2023.

The system can be broken down as follows. First, the UAV captures footage of the crowd. Then video frames from this footage are fed into the decision-making system. In this decision-making system, the features are first extracted using a ‘modified ResNet architecture.’ Then, features are selected using a ‘water cycle algorithm’ (WCA), and subsequently classified into different categories describing the level of crowdedness or the crowd behavior. Finally, this data is fed into an SMS.

The proposed model was successfully able to detect overcrowding with an accuracy of 96.55% in real-time. It was also able to detect crowd behavior, which is important for monitoring and suggesting alternative routes to prevent the spread of infectious diseases. The system is moreover robust and offers fast detection with high accuracy due to the modified ResNet architecture, which has fewer end-to-end layers.

“Our novel system can be deployed and implemented in smart cities to help to meet several social system purposes. It is a powerful tool that can help curb the spread of infectious diseases, such as COVID-19 by monitoring crowd behavior and suggesting appropriate routes for crowd movements by SMSs,” concludes Prof. Jeon.

The proposed system paves the way toward the real world application of visual sensing for the purposes of crowd control.



Title of original paper: Smart Visual Sensing for Overcrowding in COVID-19 Infected Cities Using Modified Deep Transfer Learning

Journal: IEEE Transactions on Industrial Informatics

Authors: Khosro Rezaee (1,*), Hossein Ghayoumi Zadeh (2), Chinmay Chakraborty (3), Mohammad R. Khosravi (4), and Gwanggil Jeon (5,*)


  1. Department of Biomedical Engineering, Meybod University, Iran
  2. Department of Electrical Engineering, Vali-e-Asr University of Rafsanjan, Iran
  3. Electronics and Communication Engineering Department, Birla Institute of Technology, India
  4. Department of Computer Engineering, Persian Gulf University, Iran
  5. Department of Embedded Systems Engineering, Incheon National University, South Korea

*Corresponding author’s email:,

About Incheon National University
Incheon National University (INU) is a comprehensive, student-focused university. It was founded in 1979 and given university status in 1988. One of the largest universities in South Korea, it houses nearly 14,000 students and 500 faculty members. In 2010, INU merged with Incheon City College to expand capacity and open more curricula. With its commitment to academic excellence and an unrelenting devotion to innovative research, INU offers its students real-world internship experiences. INU not only focuses on studying and learning but also strives to provide a supportive environment for students to follow their passion, grow, and, as their slogan says, be INspired.


About the Author
Dr. Gwanggil Jeon received a Ph.D. from the Department of Electronics and Computer Engineering, Hanyang University, Seoul, Korea, in 2008. Currently, he is an Assistant Professor at the Department of Embedded Systems Engineering, at Incheon National University in Korea. His research interests lie in the fields of image processing and computational intelligence, particularly in image compression, motion estimation, image enhancements, and fuzzy and rough sets theories. He is an IEEE Senior Member and has received numerous awards, including the IEEE Chester Sall Award in 2007, the ETRI Journal Paper Award in 2008, and the Industry-Academic Merit Award by the Ministry of SMEs and Startups of Korea in 2020.



Incheon National University

Academic excellence is paired with an unrelenting devotion to innovative research & global education. Link: