Ultimate & Extreme Civil Engineering

🖾 Automatic Crack Detection
_____________________________
Advanced image processing 🖼️ and 🖥 machine learning have made it possible to detect visible cracks 🖾 in the image. This technology will be helpful to the structural auditor as well as in assessing the damage to the structure.

Hello, My self is 👨‍💼 Vijay Parmar, as a structural engineer I have carried out many visual inspections 🔎 of structures 🏗️ and I always felt that I might have missed something or some points sometimes poor visibility and large-scale structures increase the number of human errors ⚠️. This made me wonder if there is any quick and accurate method to solve the issue 🤔. As a Ph.D. scholar, I thought this is the problem that need to be resolved with AI or ML.

◘ Sometimes large data may give visual blindness to the inspector and the auditor might miss some serious damage to the structure. This technology initially helps the inspector to shortlist the major defects. It will help to shorten cracks as per type, size, and causes.

◘ If we look at this tool 🛠️ from the perspective of civil engineering, auditing of country's public and private infrastructure is an impossible task to complete. And this is why most countries 🌎 have adopted RVS (Rapid Visual Screening) methods which help auditors to categorize the vulnerability of the structure on large scale (i.e. full town or village). The main disadvantage of this method is; it only covers normal issues of the building like the type of soil, the shape of the building 🏛️, the type of frame, etc. and it might miss some important issues.

◙ The research mentioned here is based on image processing with machine learning. (Machine learning is a computer program that learns and evolves with time) Here is a simple way to understand the process:
1. Photography of the site 📷 (Building 🏠, Bridge 🌉, Road 🛣️, Tunnel🕳, etc.)
2. Importing data into the computer software. 💻
3. Computer will divide the image into small parts (convolution filters)
4. These small parts will be used by a computer program to identify the probability of the crack. 🖾
5. After identification computer will detect the shape, width, and direction of the crack. (Geometry)
6. Defect data is ready for further process.

◘ In the future, these technologies will be further sharpened to detect the cause of cracks. (for ex. 45-degree crack at beam support normally caused due to failure in shear force or vertical cracks at the mid location of beam indicating excessive bending stress)

◙ If we imagine the future of structural inspection: the user needs to get a 360° image of the 🏢 building / 🕳 tunnel / 🛣️ road, Image processing will identify the members (beam/column/slab), detect the cracks with their properties (🌬️width & ➡️ direction) and automatically prepare to destress mapping of the structure (number of defect with probable causes) and also recommend repair solution.

◘ This technological advancement will help authorities to carry out large-scale audits before or after earthquakes, 🌊 floods, 🌀 cyclones, and any other natural disaster and it will help them to make preventive decisions.

~~~~~~~~~~~~~~~~~~~~~~
💠 Research Reference:
1. Imaged-based concrete crack detection... by Shengyuan Li & Xuefend Zhao
2. Automatic crack detection on road pavements... by Zhun Fan, Chong Li, Ying Chen, Jiahong wei, Giuseppe Loprencipe, Xiaopeng chen and Paula Di mascio
3. Automatic crack detection for tunnel... by Wenyu zhang, Zhenjiang Zhang, Dapeng Qi and Yun liu
4. Deep learning model for concrete recognition by Tien Thinh Lem Van-Hai nguyen and Minh Vuong Le
5. Blog on Deep learning with python for crack detection by Dimitris dais on: https://towardsdatascience.com/deep-learning-with-python-for-crack-detection-eceeeb3e182e
6. Automated 3D crack detection by Matthew M. Torok, Mani Golparvar-Frad & Kevin B kochersberger
~~~~~~~~~~
If anyone would like to dive into detail here are resources that can help you explore in detail.

💠 Prestained models (Ready made program)
1. Detect crack [Tensor flow]: https://github.com/nhorro/tensorflow-crack-classification
2. Pavement crack detection: https://github.com/fyangneil/pavement-crack-detection
3. Road crack detection: https://github.com/yhlleo/DeepSegmentor
4. Tunnel crack detection: https://github.com/shomnathsomu/crack-detection-opencv
5. Real-time (Live) crack detection: https://github.com/anishreddy3/Crack-Semantic-Segmentation
6. Bridge defect detection with AR Visualization (Photogrammetry): https://github.com/Shaggyshak/CS543_project_Image-based-Localization-of-Bridge-Defects-with-AR-Visualization
~~~~~~~~~~
✅ Copyright ©: The content of this post has a creative arrangement, annotation, and compilation of researched data and hence it is copyrighted by vijay parmar. However one can share this post with credits.

⛔ Disclaimer: This data is made available for information only. The content mentioned in this post is owned by their respective owners. Do verify the information before reacting or using it.
~~~~~~~~~~~~~~~~~~~~
TAGS




~~~~~~~~~~~
~~~~~~~~~~~~~~~~~~
📱Connect at:
Linked in: https://in.linkedin.com/in/thevijayparmar
Insta: https://www.instagram.com/civilyns
Youtube: https://www.youtube.com/c/VijayParmar
Website: https://www.thevijayparmar.com/
Telegram Channel: https://t.me/civilyns
Facebook: https://www.facebook.com/civilyns/