Case Study: Arup achieves quantitative, more accurate tunnel inspections with Smartvid.io computer vision

Using computer vision to improve infrastructure inspection & analysis

Arup, a leading global engineering firm headquartered in London, needed to improve tunnel inspection accuracy and move from subjective expert reports to quantitative measures of defects like cracks. To address this, Arup partnered with Smartvid.io to pilot computer-vision and machine-learning analysis on existing tunnel photo datasets.

Smartvid.io trained its platform on 2014 and 2016 photo sets (captured with off‑the‑shelf SLR and GoPro cameras), applied advanced vision classifiers to detect and quantify crack area, and produced a volumetric "total crack amount" metric. The Smartvid.io analysis matched human experts on detection and exceeded them in markup, analysis and quantification—demonstrating the tunnel’s health had remained the same—and enabled Arup to offer computer‑vision‑based inspection services to clients while Smartvid.io expands its defect detection to other infrastructure types.

Arup

Mike Devriendt

Associate Director