Historical_Building_Crack_2019

- An annotated benchmark image dataset for training and validation of crack detection systems based on Machine learning (ML) and, Deep Learning (DL) for a historical building.
- Real images of historical building cracks were taken at an ancient building suffering from cracking problem (the Mosque (Masjed) of Amir al-Maridani, located in Sekat Al Werdani, El-Darb El-Ahmar, in Cairo Governorate). It was built during the era of the Mamluk Sultanate of Cairo, Egypt in 1339- 40 CE. It is distinguished by its octagonal minaret and its large dome and considered as one of the most distinctively decorated historical buildings.
- Raw RGB digital images (.jpg) were captured using Canon camera (Canon EOS REBEL T3i) with 5184 × 3456 resolution over two years (2018 and 2019).
- The dataset contains most of the challenges facing historical buildings crack defect detection in real-world environments, such as dust, illumination, separators, crack-like, blurring, deep texture, wood patterns, etc.
- All images are divided into sub-images 256 X 256 to enlarge the dataset.
- The final Crack Dataset consisted of 3886 images [ 757 for crack and 3139 for non-crack]
To enlarge dataset size for training deep learning models, data augmentation process can be applied to increase training dataset size via generating new samples that are similar to the training samples. The following steps can be used:
1- Flipping image (vertically, horizontally and, vertically + horizontally),
2- Rotating image by 90 and -90 individually,
3- Flipping rotated images vertically,
4- Combining the output images of (1, 2 and 3) with the original images to create new dataset,
5- Adding noise to images of the new dataset such as Gaussian and salt and pepper noise,
6- Combining the output images of steps (4 and 5) to create the final augmented dataset.