COMPUTER VISION AND ARTIFICIAL INTELLIGENCE FOR ASSESSING SOIL SETTLEMENT IN GEOTECHNICS

Soil settlement under loading is an important geotechnical problem that can affect the performance of foundations and structures. This thesis proposes a method that combines computer vision and machine learning to assess soil-surface displacement in laboratory tests. Sequential images are captured during staged loading, reference points are tracked, and calibrated image coordinates are converted into physical displacement. A regression model may then relate settlement estimates to applied pressure, moisture content, void ratio, specimen thickness and elapsed time. Model performance can be evaluated using independent displacement measurements and the root mean square error metric. This contribution is a proposed methodology rather than a report of completed experiments; no measured results are presented.