[Edaily Reporter Shin Yeong-bin ] iParking, a parking solutions subsidiary of NHN(181710), has completed a pilot test of its artificial intelligence (AI)-based unmanned on-street parking system in Gangnam-gu, Seoul, and has begun full-scale operations.

iParking announced on the 24th that it has successfully completed the pilot test of the “Next-Generation AI Unmanned On-Street Parking System,” conducted jointly with the Gangnam-gu Urban Management Corporation, and has begun full-scale operations.

This pilot project was launched to address the limitations of existing systems installed at public on-street parking lots within Gangnam District and to validate an AI-based unmanned on-street parking operation model. The company explained that the project is significant as a collaborative model combining the public sector’s demand for administrative innovation with private-sector technology investment.

Unmanned operation was difficult for on-street parking lots without barriers, as it relies solely on license plate recognition. This is because the system must distinguish between parked and passing vehicles using only cameras and determine the occupancy status of parking spaces in real time.

Existing unmanned systems often misidentified passing vehicles as parked ones or experienced occupancy detection errors when a vehicle behind attempted to enter just as the vehicle in front was exiting. As a result, these systems remained semi-automated, requiring manual correction by staff to resolve overcharges.

iParking independently developed an AI system tailored to on-street parking environments in response to improvement requests from the Gangnam-gu Urban Management Corporation. The company implemented an integrated unmanned operating system in urban settings that functions without the intervention of on-site parking attendants—from vehicle object detection and license plate recognition to parking fee settlement.

Four core AI technologies were applied to this system. It features “Segmentation AI,” which analyzes vehicle outlines on a pixel-by-pixel basis to reduce the likelihood of false positives, and “On-Device AI,” which performs computations directly on a dedicated AI board inside the camera to minimize recognition delays.

It also incorporates “Same Vehicle Identification AI,” which combines license plate recognition with vehicle appearance similarity analysis to improve occupancy determination accuracy even in adverse weather conditions, and “Video Self-Diagnosis AI,” which automatically detects degraded video quality to preemptively prevent misrecognition. Through these features, the system aims for stable operation free of overcharges and unpaid fees.

Operational convenience has also been enhanced. Performance improvement patches and security patches are supported via automatic over-the-air updates. Dedicated monitoring software allows operators to view parking lot operational information at a glance, including real-time vehicle information, manual entry and exit processing, and usage statistics.

iParking and the Gangnam-gu Urban Management Corporation have now entered the operational phase using the actual billing system. Going forward, they plan to measure performance based on field data and continuously refine the operational model.

Starting with this project, iParking plans to expand the adoption of its system to public parking lots and on-street parking spaces nationwide.

Ha Tae-nyeon, CEO of iParking, stated, “The goal of this pilot was to verify a structure that maintains the reliability of operational data without administrator intervention, and we have successfully demonstrated this in a congested downtown setting.” He added, “Based on iParking’s technological capabilities, we will set a new standard for AI-based unmanned parking operations that spans both the private and public sectors.”