Crack Detection in Concrete Pipes Using Deep Learning Assisted Computer Vision

Water authorities spend millions of dollars unnecessarily repairing, replacing or relining reinforced precast concrete pipes that are still safe and performing to standard. Current CCTV technology used to monitor pipes doesn’t allow asset owners to distinguish between wide cracks in need of repair and narrow cracks that can self-heal. On April 14, 2026, SmartCrete CRC held this community of practice to discuss a collaborative research project developing a better way to monitor and measure pipe cracks. Chair Gavin Chadbourn (Optima Asset) chatted with project participants Yancheng Li (University of Technology Sydney) and Karen Thompson (Concrete Pipes Association Australasia) about how computer vision and deep learning can gather more precise data on crack widths to support more informed and cost-effective asset management strategies.