ABSTRACT
The growing need for efficient and safe methods for industrial asset inspection has driven the adoption of advanced technologies. This study introduces an innovative methodology that integrates Computer Vision (CV) and remote sensing techniques for structural condition assessment, tackling challenges related to data acquisition and automatic damage detection. The proposed approach comprises four key stages: (i) data acquisition using MLS, TLS and UAVs, (ii) data processing to generate a 3D model, (iii) deep learning-based damage identification using Mask R-CNN, and (iv) mapping of detected damages onto the 3D model via Ray Casting. The methodology was validated in a real-world industrial asset, demonstrating its efficacy in detecting corrosion, mechanical damage, and ponding water. The findings indicate that the proposed framework enhances automation in asset inspection, contributing to more efficient and informed maintenance strategies.
