An S-1 R&D Center researcher is demonstrating an AI agent. /Courtesy of S-1

“Show me the video of the man who collapsed 5 minutes ago on the 1st floor camera.”

The era of conversing with CCTV has arrived. While in the past, CCTV was primarily used for checking recorded footage, the intelligent CCTV 'Artificial Intelligence (AI) Agent' introduced by S-1 swiftly alerts users to emergencies or urgent situations when incidents occur, as well as guiding them on the most suitable response methods.

Additionally, users can control various functions such as video viewing and capturing through conversations with the AI Agent. This allows for resolving the operational and management issues of CCTV that have been pointed out as limitations of the existing system.

Improving the operational efficiency of CCTV monitoring has been one of the pressing issues. Although the number of installed CCTVs has rapidly increased across public institutions and private buildings nationwide, the workforce available to monitor and operate them has actually decreased.

According to the Ministry of the Interior and Safety, as of 2023, approximately 1.76 million CCTVs are installed in public institutions nationwide, which is an increase of over 1.2 million from a decade ago. However, the number of personnel directly monitoring and operating these CCTVs has declined from about 9,200 in 2011 to 4,093 in 2024. It is expected that many local governments will not meet the Ministry's standard for monitoring centers, which stipulates that there should be no more than 50 CCTVs per monitoring personnel. In this context, the intelligent CCTV 'AI Agent' is gaining attention as a new alternative to conventional monitoring methods.

S-1's 'AI Agent' offers features such as ▲SOP (Standard Operating Procedure) guidance during emergencies ▲interactive video search service ▲interactive CCTV control service. The task of having humans monitor CCTV screens 24 hours a day has inherent limitations. Long hours of monitoring can lead to lapses in attention or decreased focus due to accumulated fatigue, resulting in blind spots in monitoring. Research by the Seoul Institute indicates that monitoring efficiency decreases by nearly 95% after 22 minutes of continuous viewing of the CCTV screen.

The first step in enhancing CCTV monitoring efficiency was the commercialization of intelligent CCTV. Intelligent CCTVs analyze real-time video data to detect anomalies such as assaults, intrusions, fires, and hazardous material deposits, immediately notifying users. The AI Agent introduced by S-1 has evolved beyond existing intelligent CCTVs, autonomously detecting dangerous situations and promptly explaining them to users while suggesting countermeasures.

For instance, in the event of an accident in a warehouse, a notification saying “A man has collapsed in the warehouse” is automatically displayed on the screen. The AI then provides situation-specific SOP (Standard Operating Procedure) guidance to assist with rapid response. For example, in the case of an accident, it may guide specific action steps such as “▲Identify the accident area ▲Issue an announcement ▲Contact 119 for emergencies” directly on the screen.

Additionally, while existing CCTV systems are primarily used for securing evidence after incidents occur, searching for specific scenes can take a long time when the amount of footage is vast. If CCTV analysis takes too long following an incident, initial responses may be delayed, leading to prolonged resolution processes.

The AI Agent provides an interactive video search service to address these issues. The service's most significant feature is that it can automatically find the requested footage along with explanations if users request specific footage using voice commands. For example, if a user says, “Find the person who was smoking between 10 a.m. and 11 a.m.,” it analyzes keywords such as time and actions to quickly provide relevant footage. This function helps improve the speed of initial responses during incidents and significantly reduces the time required for CCTV analysis to facilitate prompt actions.

Most CCTV users often struggle with operating the features they desire due to unfamiliarity. The AI Agent offers an interactive control service that allows users to execute essential CCTV functions using only voice commands. For instance, simply saying, “Capture and save the front door camera screen, and show me the video from the basement entrance camera 5 minutes ago,” allows users to perform actions such as video playback, movement, and storage intuitively.