Crack Monitoring from Motion (CMfM) integrates photogrammetric techniques with deep learning methods for the automatic detection and monitoring of cracks. CMfM uses a series of images collected by non-fixed cameras to monitor crack propagation over time. Unlike conventional techniques, CMfM does not require fixed artificial targets and overcomes the limitations of using a fixed camera of the 2D DIC.
The approach employs Convolutional Neural Networks (CNNs) for automatically detecting the shape of the cracks and a skeletonization approach for delineating the centre line of the defects and for automatically selecting the points of interest around the cracked area. Then, it computes the change in the distance between the selected points on the left and right sides of the crack for estimating the crack width propagation over time. The proposed approach is based on homography estimation and template matching techniques. The workflow is shown below.
This work is part of the TACK project.
To use the code on your local machine, use the following instructions:
- Download Python 3
- Install the packages:
pip install -r requirements.txt
To run the code on Google Colab, you can run the notebook.
Datasets related to the project:
If you use the datasets in your research, please cite:
- Andreas Sjölander, Valeria Belloni, Viktor Peterson and Jonatan Ledin (2023). Experimental dataset to assess the structural performance of cracked reinforced concrete using Digital Image Correlation techniques with fixed and moving cameras. In: Data in Brief, Volume 51, https://doi.org/10.1016/j.dib.2023.109703
- Andreas Sjölander, Valeria Belloni and Andrea Nascetti (2022), Dataset to track concrete cracking using DIC with fixed and moving camera, Mendeley Data, V1, doi: 10.17632/dns97tfdjn.1
- Andreas Sjölander, Valeria Belloni, Viktor Peterson and Jonatan Ledin (2023), Dataset to assess the structural performance of cracked reinforced concrete using FEM, DIC and CMfM, Mendeley Data, V2, doi: 10.17632/z3yc9z84tk.2
If you use CMfM in your research, please cite the following papers:
-
Valeria Belloni, Andreas Sjölander, Roberta Ravanelli, Mattia Crespi and Andrea Nascetti (2023). Crack Monitoring from Motion (CMfM): Crack detection and measurement using cameras with non-fixed positions. In: Automation in Construction, vol 156, https://doi.org/10.1016/j.autcon.2023.105072
-
Valeria Belloni, Nuhamin Deresse, Andrea Nascetti and Els Verstrynge (2025): Crack monitoring of masonry walls with standard and enhanced Digital Image Correlation methods, 6th Joint International Symposium on Deformation Monitoring (JISDM 2025), Karlsruhe, 7-9 April 2025, https://doi.org/10.5445/IR/1000180531
For general information on the TACK project, please refer to:
- Valeria Belloni, Andreas Sjölander, Roberta Ravanelli, Mattia Crespi and Andrea Nascetti (2020). TACK PROJECT: TUNNEL AND BRIDGE AUTOMATIC CRACK MONITORING USING DEEP LEARNING AND PHOTOGRAMMETRY. In: Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLIII-B4-2020, 741–745, https://doi.org/10.5194/isprs-archives-XLIII-B4-2020-741-2020
For general information on the use of standard Digital Image Correlation (DIC), please refer to:
- Valeria Belloni, Roberta Ravanelli, Andrea Nascetti, Martina Di Rita, Dimitilla Mattei and Mattia Crespi (2019). py2DIC: A New Free and Open Source Software for Displacement and Strain Measurements in the Field of Experimental Mechanics. In: Sensors 19, 19, 3823, https://doi.org/10.3390/s19183832
Code is released for non-commercial and research purposes only. For commercial purposes, please contact the authors.
To contribute or ask for information contact [email protected].