Preventing disasters such as large-scale fires and natural disasters is more socially and economically cost-effective than recovering from actual incidents or accidents that cause damage. Recent trends in disaster safety technology are increasingly embracing vision AI, a promising technology that reduces human resources and human error while also offering excellent compatibility and integration with existing infrastructure.

The Korea Safety Industry Expo, which was held for three days from the 13th, was filled with various safety solution products to prepare for industrial accidents such as worker entrapment accidents in manufacturing plants and falls at construction sites, as well as disaster situations such as fires and accidents.

■ Utilizing AI for safety in manufacturing and construction sites

With the introduction of the Serious Disaster Punishment Act and the resulting increased need for industrial accident prevention, the safety industry is gaining momentum in adopting new technologies. Among these, vision AI technologies and products utilizing them are proving to be a major focus, attracting the attention of visitors.

InterX is a specialized industrial AI service company with a history of participating in various smart factory construction projects and development in the field of vision AI inspection. At the exhibition that day, InterX unveiled 'Safety.AI', a solution for ensuring the safety of field workers and preventing major accidents.

InterX's 'Safety.AI' is an intelligent CCTV product that can monitor and notify of dangers such as △safety check of dangerous areas △preemptive collision prevention △fall detection △entrapment prevention. It comes with a built-in kiosk hardware that can analyze up to 4 channels of video in real time, so it has the advantage of being ready for use without building an additional system.

Small businesses tend to be reluctant to spend large amounts of money on safety systems, but thanks to lower consumer prices through support projects from the Korea Occupational Safety and Health Agency, it has become relatively easy for small and medium-sized businesses to introduce safety systems for specific areas.

Seong In-chae, head of the AI business development team at InterX, explained, “In the safety industry, vision AI is just beginning to take shape, and customers do not have a high need for safety unless there are special issues, as it is a cost unrelated to production.” He added, “When regulations such as the Serious Disaster Prevention Act are enacted, market interest increases, so safety solutions such as vision AI will benefit from regulations and laws.”

■ Tunnel Crack Inspection, Instantly Solved with Vision AI

Large-scale infrastructure such as bridges, dams, and tunnels are essential for national and industrial development, but they also form an industry in their own right. These infrastructures not only require construction, but also require significant human, material, and time costs to maintain, manage, and repair them.

In the safety industry, especially, inspecting infrastructure for cracks is a routine task. Traditionally, inspections require significant human resources and time. This process often faces challenges related to human error and efficiency. Vision AI-powered inspections, similar to the benefits offered by quality inspections in manufacturing, can offer significant human, material, and time-saving benefits in infrastructure safety inspections.

Deep Inspection, a venture company specializing in AI safety inspections, developed the "Tunnel Deep Scanner" product through a research project funded by the Ministry of Land, Infrastructure and Transport. An optical camera installed inside a truck captures images of tunnels and roads, and AI analyzes the images to determine the extent of cracks.

It has the advantage of being able to perform inspections more than five times faster than visual inspections and preventing safety accidents for inspectors. Previous inspections required road control and inspectors to walk inside tunnels to check for safety, but inspections have become relatively easier with the ability to obtain high-resolution images through lights and cameras installed in vehicles.

Kim In-soo, Managing Director of Deep Inspection, emphasized, “The civil engineering field has a conservative atmosphere when it comes to introducing new technologies, but as we have developed advanced technologies and accumulated performance since 2019, there are now many opinions in the field that we should replace them with this method.” He added, “We have introduced the concept of eXplainable AI (XAI) and introduced technology that expresses the process of how the results were obtained in text.”

Recently, many countries, including the US and the EU, have set explainability of AI as a prerequisite for the safe introduction of AI. Therefore, whether or not a company providing the same AI model satisfies the XAI standard will likely become a key factor in future market competition.