An rp2040-based acquisition board for pulse-echo ultrasound

Incorporating Inductive Bias into Deep Learning: A Perspective from Automated Visual Inspection in Aircraft Maintenance

Application of LSTM network for Structural Health Monitoring & Non-Destructive Testing

Official Python implementation for XVis Toolbox release with the book Computer Vision for X-Ray Testing.

GAN/convolutional and Transformer models to predict missing mechanical information given limited known data in part of the domain, and further characterize the composite geometries from the recovered mechanical fields for 2D and 3D complex microstructures

Ultrasonic testing (UT) is a family of non-destructive testing techniques based on the propagation of ultrasonic waves in the object or material tested.

StofNet: Super-resolution Time of Flight Network (ICASSP 2024)

Ultrasonic NDT For Surface Fault Detection

Paper: A new ML-based model for automatic flaw detection in pipe inspections using an acoustic lens.

Paper: An acoustic lens that enables ultrasonic thickness measurement of large pipe sections.

A MATLAB implementation for non-destructive testing of heterogeneous materials using ultrasound. This project simulates wave scattering from structural defects and applies the Differential Linear Sampling Method to detect and localize anomalies such as cracks.

Hardware source (Altium) from my bachelor's thesis in telecommunications engineering at the University of Granada, 2022.

GNN-based CFRP defect localization framework — companion code for Frontiers in Materials 2025 (DOI: 10.3389/fmats.2025.1652484)

FMC amplitude reconstruction from binary ultrasonic data with a lightweight MobileViT-v3-v1-FPN-PixelShuffle

Implements MATLAB simulations from the paper "Qualitative indicator functions for imaging crack networks using acoustic waves". Includes two inversion algorithms for crack detection based on far-field acoustic data.

Methods for the automated detection of acoustic multiplets and their hierarchical classification according to the similarity of their emission mechanisms

HARD, NDT: A post-processing pipeline mapping relative hollowness by fusing audio transients with computer vision. It analyzes spectral resonance decay against a solid baseline to detect density shifts. Uses OpenCV for spatial tracking and renders 3D topographic heatmaps via PyVista.