Yongsan-gu deploys AI CCTV network to monitor Halloween crowds in Itaewon

Seoul's Yongsan-gu is mobilizing all available resources for Halloween crowd management as the fourth anniversary of the Itaewon tragedy approaches.

The district has deployed AI-powered smart closed-circuit television cameras to prevent a repeat of the crowd crush.

District Mayor Kim Kyung-dae has been personally visiting sites throughout the Halloween special-measures period to review safety operations. From 10 p.m. Friday, Kim toured the district's disaster safety operations center, the surveillance control center and the Itaewon World Food Culture Street to check real-time crowd management conditions and assess on-site response systems.

A total of 218 AI smart CCTV cameras are operating at 59 locations in and around Itaewon where large crowds are expected, continuously analyzing congestion levels in real time. When warning signs are detected, alerts are immediately relayed to the district's disaster safety operations center and to police and fire department units so that a rapid on-site response can follow.

AI smart CCTV cameras have also been installed along major commercial strips and high-density areas — including Gyeongnidan-gil, Yongni-dan-gil and Sinheung Market, as well as World Food Culture Street — to monitor crowd flow and congestion conditions across each zone in real time.

The AI smart CCTV system automatically counts the number of people in each camera's field of view and analyzes crowd density per unit area in real time. Moving beyond the conventional approach of having control-room operators manually scan dozens of screens, the system quantifies crowd growth and density patterns and flags danger signs quickly.

Analyzed crowd-density data are automatically classified according to a four-tier congestion standard. When the danger threshold is reached, footage from the affected area is prioritized on the control-room display and alerts are immediately sent to the district's disaster safety operations center and to police and fire department units. Three electronic information boards along Itaewon World Food Culture Street display real-time crowd information and evacuation directions in both Korean and English so that visitors can respond to congestion on the spot.

Several additional data sources are integrated to sharpen crowd-density assessments. The district combines AI CCTV video analysis with the "Yongsan Real-Time Smart Map" — a system that collects pedestrian flow data every five minutes using mobile carrier base-station information — along with exit counts from Itaewon Station and the Seoul Metropolitan Government's real-time urban data platform. A double-check mechanism is also in place: whenever a crowd-risk alert arrives from the Ministry of Interior and Safety's mass-gathering notification system, operators verify conditions on the ground through AI CCTV footage.

During the Halloween special-measures period running from Oct. 23 through Nov. 1, AI CCTV cameras are under intensive monitoring to track crowd movement closely. If warning signs are detected, a joint on-site command post and safety personnel deploy immediately to disperse and control the crowd. The district plans to keep the AI crowd-management system running year-round for other large gatherings, including year-end and New Year events.

The district is also stepping up field-based safety patrols. Starting at 7 p.m. Friday, Yongsan-gu conducted a joint civilian-government-police night patrol across the Itaewon area in anticipation of rising Halloween visitor numbers. About 40 people took part — including officials from Yongsan-gu, Yongsan Police Station, the Yongsan-gu Autonomous Crime Prevention Federation and neighborhood crime-prevention units — focusing on World Food Culture Street, Vietnam Quinon-gil and the area around Itaewon Station.

The patrols were used to identify potential crime and safety hazards in crowded areas in advance, with any on-site risks addressed immediately. Officers also carried out outreach activities to protect vulnerable residents, including guiding minors and helping elderly residents with dementia find their way home.

"A single moment of delayed judgment in a crowd incident can lead to irreversible consequences," District Mayor Kim said. "We will use our AI-based surveillance system to detect warning signs proactively and operate a seamless response chain that moves instantly from situation detection to on-site crowd control and dispersal." Kim added that the district would "deploy every advanced technology and field-response capability to protect the safety of residents and ensure not a single safety incident occurs."