A team from Drexel University developed a novel machine learning-based system to enhance visual inspection of infrastructure, aiming to tackle the challenges of aging structures and limited inspection resources. Combining computer vision with deep learning, their multi-scale approach identifies and assesses problem areas, streamlining inspection processes. 

By utilising stereo-depth cameras and robotic scanning, the system detects and measures cracks with exceptional precision, outperforming conventional methods. 

This innovation promises to significantly reduce inspection workload and enhance maintenance efforts, vital for preventing structural failures.

The team envisions integrating this technology into autonomous monitoring frameworks, advancing infrastructure maintenance towards a more comprehensive and efficient approach.

Read the original source