A Proposal for the Utilization of Sensors for Crowd Monitoring in Shopping Malls in Singapore - Draft 1

  1. Introduction

This proposal was prepared in response to the request for proposals on developing solutions for the engineering problem that is the usage of sensors for automated crowd monitoring in commercialized buildings in Singapore. It aims to provide building management with a focused overview of how such measures will bring benefits.

According to Coolfire Core (2019), crowd monitoring refers to the ability to control, manage and monitor large groups of people to ensure the safety and efficiency of people in case of emergencies, such as fire, terrorist attacks, etc. Modern technology, engineering, and psychology work together in the implementation of automated crowd monitoring. Monitoring crowds has gradually become important nowadays due to the growing population in Singapore and increasing crowd sizes. An automated crowd monitoring should have the ability to identify and track different and multiple objects as well as to recognize human subjects (Kang et al., 2019).

In light of the COVID-19, the government has established policies and measures such as screening stations, contact tracing (e.g: SafeEntry), vaccination to control and prevent the transmission of COVID-19. Crowd monitoring can be as important as temperature taking at the entrances of the malls in order to ensure safe-distancing and compliance with policies (NTU, n.d.). 

In Singapore, crowd monitoring is not implemented with the utilization of sensors broadly and still relies on manpower. A company in Japan, Optex Co., Ltd. (2021) has introduced Crowd Alert Sensor, which is a passive infrared sensor to detect the number of people passing and inform the crowd status to the visitor using built-in indicator light. In this report, we propose a solution to automated crowd monitoring contributing to both normalcy and COVID-19 situations by implementing up-down counters and infrared sensors and providing the information of crowd status via a phone application so that the public can check the crowd status even before reaching the malls.

  

1.1 Literature Review

We will look into the existing or proposed solutions done by the researchers from different countries in this section.

1.1.1

Kang et al. (2019) conducted controlled tests in indoor areas (e.g. shopping malls) and non-controlled tests in outdoor areas (public transport hub) using both methods of Background Subtraction (BGS) and Single Shot Detector (SSD), which involve video analytics. In BGS, levels of brightness and colour will be corrected to adjust varying light conditions. A known background frame with no present object will be taken as a reference, and the background reference is then subtracted from each frame of the video footage. However, the limitations in BGS are weather conditions, illumination changes, reflections from surfaces on moving objects. In SSD, it runs the full incoming frames through Convolutional Neural Network (CNN) and yields bounding boxes and class probabilities on identified objects by matching the appropriate anchor box (Kang et al., 2019).

Kang et al. (2019) concluded that SSD shows better accuracy and results in both indoor and outdoor areas due to its performance in multiple-subject scenarios. One of the reasons that BGS does not work well is because of the rapidly varying background at outdoor areas.

1.1.2

Students of Yunnan University use Wi-Fi signals to determine and analyse crowd behaviour of a science museum on campus (Chen et al., 2020). The increasing usage of mobile Wi-Fi terminals allowed the student researchers to come up with ways to improve management capabilities on campus. The students use the Media Access Control address (MAC) and Wireless Access Point (WP) to study the said problem.

The Wi-Fi monitoring apex comprises a Wi-Fi probe, computer, storage, processing and prediction module and the wireless bridge. The information from the MAC and WP of the user using the Wi-Fi is captured by the Wi-Fi probe frames. This information is then sent to storage using a non-relational database. Data Processing and Prediction is done by using a python specific computing module and a long short-term memory algorithm respectively. Lastly, the visualization of the data is developed using Django, a web framework and is connected to a cloud or server using the wireless bridge (Chen et al., 2020).

1.1.3

Determe et al. (2020) has implemented a crowd monitoring system based on probed requests, which are Wi-Fi packets that are internal to the Wi-Fi protocol and broadcasted by user terminals such as a smartphone or a computer to detect nearby access points. It is also suited to events that cameras cannot easily monitor, such as those with a complex setup that entails line-of-sight obstructions or poor lightning conditions; conventional cameras could also provide imprecise counts for open-air events because of weather effects (e.g., heavy rain and fog).

1.1.4 IR Sensor

According to Hu et al (2018), infrared detection is the use of infrared detectors and optical imaging objective to accept the infrared radiation energy distribution of the target reflected to the infrared detector photosensitive element to obtain the object of infrared thermography. Its advantages are uninterruptible power supply, non-sampling, non-contact, operational safety, etc., and real-time, fast and accurate monitoring of the operating staff status.

Based on the above considerations, Hu et al (2018) designed a monitoring system of electric business hall based on infrared imaging, which can quickly and accurately monitor the operating status of the business hall and make corresponding decisions by using a monitoring system. The experimental results show that monitoring system of electric business hall based on infrared detection can improve the system testing accuracy, improve the operating efficiency of the business hall and help to reduce the operating and maintenance costs.

Hu et al (2018) has presented an infrared thermal imaging based on the power monitoring system, through analyzing the service business process of the electric business hall. Monitoring system consists of infrared system equipment, centralized monitoring platform. Infrared system as an independent sensor system to monitor the flow of business staff, to provide clear imaging and temperature measurement, and accurate grasp of the situation at the scene. At the same time, it uploads to the centralized monitoring platform by using Transmission Control Protocol/Internet Protocol (TCP/IP) communications. Monitoring platform real-time display through data and images, and it accurately and directly understands and grasps the various business halls.

Hu et al (2018) has concluded that in view of the disadvantages of traditional video monitoring system such as inaccurate monitoring, inefficient monitoring, and inability to achieve real-time monitoring of the staff, this paper researched and designed a monitoring system of electric business hall based on infrared detection. The monitoring system uses infrared imaging technology to design a set of monitoring software of electric business hall, which achieved accurate tracking and supervision of staff and real-time control of user-side conditions. The experimental results show that compared with the traditional video technology, the monitoring system of electric business hall based on infrared imaging has good electromagnetic compatibility and electrical isolation performance. This technology reliability strongly, and the accuracy of equipment by more than 98% under the condition of not being affected by light in practical application.

1.2 Problem Statement

Ideally, automated crowd monitoring in shopping malls should be conducted to reduce the congestion in the malls so as to improve the safety measures, eliminate the risk of human error and manpower and reduce the risk of transmission in response to COVID-19 in the malls. Up-down counters will be implemented with infrared sensors at the entrances and exits of the malls to detect people passing. The crowd status will be shown on a phone application so that the public can check the crowd status before entering the malls.

1.3 Purpose Statement

This report proposes to the management of Jurong Point, Mercatus, to install up-down counters and infrared sensors at the entrances and the exits of the malls to detect the people passing as well as to implement the use of phone applications to show the crowd status to the public, so the safety measures can be improved.

References

Chen, L. W., Chen, Z. G., Huang, C. X., Lu, Y. Q., Yang, J. J., & Huang, M. (2020, October 2020). Study of the Crowd Behavior in Campus Based on WIFI Probe and Time Series Analysis. 2020 XXXIIIrd General Assembly and Scientific Symposium of the International Union of Radio Science. https://doi.org/10.23919/ursigass49373.2020.9232029

https://ieeexplore.ieee.org.singaporetech.remotexs.co/document/9232029

Coolfire Core. (2019, February 19). How To Improve Your Crowd Control Strategy With Smart Crowd Monitoring. Coolfire Blog. 

https://www.coolfiresolutions.com/blog/crowd-control-smart-crowd-monitoring/

Determe, J.-F., Singh, U. Horlin, F., & Doncker, P.D. (2020, May 12). Forecasting Crowd Counts With Wi-Fi Systems: Univariate, Non-Seasonal Models. IEEE Transactions on Intelligent Transportation Systems, vol. 22, no. 10, pp. 6407-6419, Oct. 2021, doi: 10.1109/TITS.2020.2992101. https://ieeexplore.ieee.org.singaporetech.remotexs.co/document/9091918 

Hu, B., Liu, H., Liu, J., Li, M., & Wang, Y. (2018, July 9). Design of electric business hall monitoring system based on infrared imaging. 2018 Chinese Control And Decision Conference (CCDC), 2018, pp. 5553-5557, doi: 10.1109/CCDC.2018.8408099.

https://ieeexplore.ieee.org.singaporetech.remotexs.co/document/8408099

Kang, H. C., Poeschmann, S., Lai, J. W., Koh, J. M., Acharya, U. R., Yu, S. C. M., Tang, K. J. W. (2019, December 6). Practical Automated Video Analytics for Crowd Monitoring and Counting. IEEE Access, vol. 7, pp. 183252-183261, 2019, doi: 10.1109/ACCESS.2019.2958255 

https://ieeexplore.ieee.org/document/8926351

Nanyang Technology University (NTU). (n.d.). AI-based Video Analytics for Pandemic Management. https://www.ntu.edu.sg/rose/research-focus/deep-learning-video-analytics/ai-based-video-analytics-for-pandemic-management