As CCTV installations increase and their functionalities advance, there is growing interest in protecting video information, particularly regarding personal data within footage. VECTORSIS has developed a software platform that combines digital twin and visual AI technologies, enabling real-time monitoring without personal data exposure by visualizing people as icons, avatars, or skeletons. Taehoon Kang, the founder of VECTORSIS, has been a veteran developer since his college days, working on 3D visualization projects across various fields, including geographic information systems, robotics simulations, gaming, and healthcare. We spoke with CEO Kang to learn more about VECTORSIS as a company and how digital twin technology integrates with video security and surveillance systems.
What kind of company is VECTORSIS?
Founded as a one-person company in 2017, VECTORSIS expanded its team in 2020, graduated from SK Telecom’s accelerator program, and secured an initial investment of 500 million KRW, establishing itself as a promising SME in Gyeonggi Province. As an experienced developer in 3D visualization projects across GIS, robotics simulation, gaming, and healthcare, I created a royalty-based business model by licensing our surveillance software developed in the early days of the company.
With the initial investment, we developed a software platform that integrates digital twin and visual AI technologies. Based on this, our XR Twin System (XRTS) replicates real spaces and objects in a digital space to enable video security and automated control. Following a successful demonstration project last year, XRTS is now in the commercialization phase.
What led you to integrate 3D digital twin technology into video surveillance systems?
Traditional video security and surveillance systems display camera feeds in a grid format, making it difficult to understand the actual structure of a building and identify spatial relationships between feeds. The idea of integrating 3D digital twin technology into video surveillance systems emerged to address these issues. We developed software that performs patrol and access control functions in a virtual space, which has been implemented at over 180 sites through partner companies.
What makes VECTORSIS’s intelligent video surveillance and analysis system competitive?
From a technical perspective, instead of using game engines like Unity to implement the digital twin, we utilize a custom-developed 3D rendering engine. This allows us to tailor visualization processes such as projection mapping, which applies camera footage to virtual buildings, and optimize the rendering pipeline to run on lower-spec systems, reducing product costs.
In terms of functionality, our system visualizes AI-detected people or vehicles as icons, avatars, or skeletons rather than actual footage, enabling real-time monitoring of diverse sites while protecting personal privacy.
What is the XR Twin System (XRTS)?
XRTS is a platform that synchronizes real-world data from CCTV cameras, HVAC systems, and industrial machinery with 3D virtual structures, allowing real-time monitoring of urban infrastructure such as buildings, factories, roads, ports, and railways. Unlike other digital twins that only sync information received from control devices, XRTS visualizes detected objects such as people, vehicles, and robots from camera feeds, enabling comprehensive real-time monitoring of all objects and conditions within physical spaces. Earlier this year, XRTS passed the Personal Information Protection Commission’s preliminary feasibility review, allowing installation in sensitive work sites where CCTV is typically restricted.
XRTS also serves as an intelligent video surveillance platform that analyzes behavior to detect abnormal situations such as falls, intrusions, wandering, or entrapment and actively issues alerts. Additionally, XRTS offers a plugin-based design, enabling technical support and joint development to transform other companies' video surveillance displays into digital twins.
What does passing the preliminary feasibility review mean?
The preliminary feasibility review system is designed to help businesses in new service and technology fields comply with the Personal Information Protection Act by working with the Personal Information Protection Commission (PIPC) to establish appropriate privacy measures. If the business applies these measures correctly, it can avoid administrative penalties as long as circumstances remain unchanged.
Simply put, this means we can install our XRTS in areas with video surveillance without requiring consent from data subjects. This is based on XRTS’s ability to display and record only incident scenes, exclude storage of routine footage, replace individuals in video feeds with icons, and transparently disclose operating policies to data subjects within the surveillance area. The PIPC concluded that the legal benefits of preventing and analyzing incidents outweigh any concerns of employee monitoring.
Could you share an example of an XRTS deployment?
Last year, XRTS was applied to a demonstration project on the training vessel of Korea Maritime and Ocean University in Busan, where it detected falls and other incidents in real-time. During this project, an AI model was developed to detect objects in thermal camera footage, and pose estimation was applied for more accurate behavioral analysis. We chose the training vessel as a demonstration model because, given the ship’s size, incidents like falls often go unnoticed. Ships are also suitable for 3D modeling, and precise location tracking is crucial in fall detection, so we conducted the demonstration on this vessel. Currently, we’re implementing XRTS in manufacturing sites to detect incidents like falls, collisions, and entrapments in real-time. We are also working with SK affiliates to integrate XRTS into surveillance systems for experience parks and automated control systems at U.S. military bases.
You mentioned that XRTS has four different display modes for privacy protection. Could you explain each one in detail?
XRTS uses four display methods: Face Masking, Skeleton, Pose Estimation, and Digital Twin.
- Face Masking: This method mosaics faces in real-time. However, there is a risk of exposure if the AI fails to detect a face.
- Skeleton: This method replaces individuals with skeletal figures detected by AI. It prevents personal identification and allows differentiation of the head, body, arms, and legs using colors.
- Pose Estimation: This mode removes the background and displays individuals as skeletal figures, ideal for high-security areas.
- Digital Twin: This method not only represents individuals as skeletal figures but also builds the monitored area as a 3D virtual space. However, it requires additional costs to establish the digital twin space.
What challenges do you face in developing AI solutions and products?
At VECTORSIS, we use AI as a means to build digital twins. A key challenge is developing new technologies that align with existing regulations, which helps reduce obstacles at the commercialization stage. Due to regulatory constraints, additional time and costs are sometimes required to create alternative solutions, which can affect the accuracy of the technology.
For example, to ensure precise tracking, facial recognition would be ideal. However, current laws prohibit the storage of facial data. It would be helpful if facial information could be used for tracking in urgent situations, such as locating suspects or missing persons.
What are VECTORSIS's future plans and goals?
Numerous CCTV cameras are already in place, supporting various solutions like incident detection, parking management, and suspect tracking. However, only certain authorized individuals, like government and security personnel, have access to this footage. Due to privacy laws, real-time video isn’t available to the general public, except in some areas on highways. VECTORSIS aims to develop and distribute a platform that analyzes camera footage and provides useful information to the public in real-time, all while adhering to relevant laws.
Read the full article
