Guided wave testing (GWT) is a non-destructive testing (NDT) technique for in-service testing of pipes allowing the inspection of tens of metres of pipe in either direction from a single position. The aims are to identify and locate physical features along the pipe in the axial direction, particularly the defect indications, such as cracks or corrosion patches. However, the signals output by GWT of pipes are complex to interpret, making the quality of inspection highly dependent on the operator’s skills. Due to signal complexities, at present there are no automated procedures to help operators in this task. Some of the recently developed machine learning (ML) algorithms are expected to possess the modelling capabilities required to address this classification task, though they would typically need hundreds if not thousands of labelled input data for their training. This amount of experimental data is seldom available in the NDT field, particularly with regard to the damage cases. This thesis explores the ML for NDT, introducing the data processing pipelines and a comparison of ML approaches. First, it is shown that high ML performance on artificial data does not necessarily translate to a similar performance on real data, motivating the need for robust ML model validation. The following results demonstrate that with scarce experimental data, substantial detection improvements can be achieved by pre-training the chosen ML model with synthetic data, before fine-tuning it on actual inspection data. In particular, the ML algorithm that is found to perform best for this task is a VGG-Net model, which is shown to yield false positive rates in the order of ~1.5 to 4% at the fixed true positive rate of 99.7%. Furthermore, the thesis explores modern generative ML approaches as potential tools to augment the data, showing the capacity to generate realistic data ex nihilo.