SEOUL, SOUTH KOREA, June 24, 2021 – Seoul becomes the first city in Korea to use AI technologies to identify sewer defects, such as holes and collapse. So far, the pipeline inspection has been done manually by public officials.
Last year, the Seoul Digital Foundation (SDF) developed the “AI-based sewer defect detection system.” The system automatically identifies problems of sewers from CCTV images. The SDF devised this system using computer vision that recognizes defects in pipelines through digital images and videos.
Sewer inspection has long been manually conducted by public officials. However, as there are around 9,000km of public sewers across Seoul, detecting damages to the circular sewers with naked eyes was incredibly taxing.
From now on, AI will automatically sort the type and location of defects in the sewer by interpreting CCTV images, then shows the problematic location on the inspector’s screen. To do this, the SDF categorized ten types of sewer defects and trained the AI with 5,000 CCTV image data over the last year.
The Seoul Metropolitan Government expects that once the “AI-based sewer defect detection system” is introduced, the accuracy will increase while the time consumed in the task will decrease, allowing the city to prevent accidents in advance.
In April, the SDF started to run pilot programs of the system in two districts of Seoul. At the same time, the foundation is conducting an additional study to categorize the defects in19 types and further improve the accuracy of the AI sewer inspection. By applying the results of pilot programs and the research, it aims to expand the system to every district next year.
Also, the SDF is going to run its “data for detecting changes in urban buildings through aerial videos” project from this month. The new project uses AI in recognizing changes in urban buildings, such as construction and dismantlement of buildings. Based on aerial videos of Seoul, the project will secure 50,000 images for AI training and develop AI models that automatically detect changes in structures.
Just like the sewer inspection, detecting changes in buildings was manually done by experts. But once AI technologies are introduced, the task will be automated, requiring a smaller amount of time. In addition, the accuracy of the job will rise, enabling the city to spot unauthorized buildings and rapidly respond to collapse and fire.
Lee Weon-Mok, Acting President of the Seoul Digital Foundation, mentioned, “We will further improve the quality of our AI-based innovative administrative service. To that end, we will advance current technologies to a practical level by reflecting the voices of officials on site.”