Published · 2020 and 2022
Read structural condition from images and signals.
Connect crack classification and measurement with vibration-based damage localization.
Portfolio revision: 17 October 2026
Research question
How can learned image and signal features support inspection and damage localization?
Method & evidence
The image study combines convolutional classification with image processing and geometric measurement. BuildingNet learns features from acceleration windows and examines channel count and noise under numerical and experimental cases.
Finding & status
The published crack study reports a 96.17% classification test accuracy for its combined classification experiment. BuildingNet studies single-channel detection and localization under its defined structural cases.
Conditions of interpretation
The classification score belongs to the crack article’s test cohort. It is separate from geometric-measurement error and from the later BuildingNet validation.
My contribution
I authored the related Western master’s thesis and am first author of the crack-assessment and BuildingNet articles. The thesis documents my image and vibration research programme.
Authorship
Coauthored studies; the 2020 thesis is an independently identified companion methods source.
Data & project materials
The Western repository provides the master’s thesis. The journal records provide article and supplementary-material access routes.