This project include several different surfaces, each surface contains one or several defects. For segmentation,object detection, saliency detection，etc
Researchers of surface defect often suffer from the lack of the corresponding dataset.
Thanks to the data provider's generous support, we can collect these datasets.
You can download some of these datasets which are modified by us, they are in the zip files in this project. Others datasets you can visit the links we provide.
The image datasets are only for academic research, no commercial purposes are allowed. If you use any datasets, please cite the paper of the corresponding provider
There are several different surfaces, each surface contains one or several defects. For image segmentation, object detection, saliency detection, classification, etc.
The image datasets are:
Magnetic tile dataset by ourselves. Can be downloaded from https://github.com/abin24/Magnetic-tile-defect-datasets. which was used in our paper "Surface defect saliency of magnetic tile", the paper can be reach by here or here
these three datasets are used in our paper A Compact Convolutional Neural Network for Surface Defect Inspection
Cracks on the bridge(left) and crack on the road surface.
Bridge cracks. There are 2688 images of bridge crack without pixel-level ground truth. From the authors "Liangfu Li Weifei Ma Li Li Xiaoxiao Gao". Files can be reached by visiting https://github.com/maweifei/BridgeCrackImage_Data.
Crack on road surface. From Shi, Yong, and Cui, Limeng and Qi, Zhiquan and Meng, Fan and Chen, Zhensong. Original dataset can be reached at https://github.com/cuilimeng/CrackForest-dataset. We extract the image files of the pixel level ground truth.
These Datasets from Kechen Song, Northeastern University (NEU). You can visit their Homepage http://faculty.neu.edu.cn/me/songkc/Vision-basedSISSteel.html to download 1800 images.
-examples of this dataset are shown in the figure below:
You can download these 10 datasets by visit https://hci.iwr.uni-heidelberg.de/node/3616 Samples like the figure below: