Vision-based crack detection with crack propagation modeling

In this work, the propagation of cracks on some materials, such as concrete, is analyzed, the best-fit model of which is employed for vision-based crack detection. Some research questions are being investigated:

• 1. What is the correlation between simulated crack propagation and real-world crack? Here, we measure the fitting errors between best-fit regression models of simulated cracks and the skeleton of crack images from some reputable image datasets.
• 2. How can we leverage the information obtained from simulations to enhance the performance of crack detection models? Some approaches are being conducted: i) employing the propagation similarity as a feature to develop post-processing classifiers, ii) generating GAN-based quality training samples using crack models and background pixels from other reputable crack image datasets.

Interested research students are welcome to join. Minimum requirements: coding skills (MATLAB, Python for algorithm development and implementation), English reading and writing (to read papers and write reports, for now. Students are encouraged to strengthen their verbal skills as soon as possible).

Selected Publications

T. H. Dinh, V. T. T. Anh, T. Nguyen, C. H. Le, N. L. Trung, N. D. Duc, and C.-T. Lin, Toward vision-based concrete crack detection: Automatic simulation of real-world cracks,. IEEE Transactions on Instrumentation and Measurement, vol. 72, pp. 1–15, 2023.

T. H. Dinh, C. H. Le, Q. P. Ha, UAV Imaging: Correlation between Contrast and F1-score for Vision-based Crack Detection. The 16th IEEE/SICE International Symposium on System Integration (SII 2024), Ha Long, Vietnam 8-11th January 2024.

Other information

Co-PI: Dr. Dinh Tran Hiep