Determination of Internal Elevation Fluctuation from CCTV Footage of Sanitary Sewers Using Deep Learning

The slope of sewer pipes is a major factor for transporting sewage at designed flow rates. However, the gradient inside pipe changes locally various reasons after construction. This causes disturbances requiring investigation and appropriate maintenance. study extracted internal elevation fluctuation from closed-circuit television footage, which required sanitary sewers. principle that change in water level indirectly indicates was applied. area detected using convolutional neural network, type deep learning technique, calculated geometric principles circles proportions. training accuracy 98%, compared to random sampling 90.4%. Lateral connections, joints, outliers were removed, smoothing method applied reduce data fluctuations. Because target are 2.5 m concrete reinforced pipes, joint determined every so would consist linear slopes. investigative proposed this effective with high economic feasibility sufficient existing sensor-based methods investigation.