Fabric Defects Dataset
Description
The data collected for this project were from : ● Fabric Defect Dataset from Kaggle (Ranathunga, 2020) ● Fabric Stain Dataset from Kaggle (Pathirana, 2020). ● Aitex fabric image database (Silvestre-Blanes et al., 2019) ● Dataset from the author of the literature review paper (Peng, 2020): only the non-defect images from the dataset. Aitex fabric image database This textile fabric dataset consisted of 245 images of 7 different fabrics with image sizes of 4096×256 pixels. There were 140 defect-free images, 20 for each type of fabric. In each of the defected category, there were 105 images (Silvestre-Blanes et al., 2019). Fabric Stain Dataset This was taken from Kaggle (Pathirana, 2020). The dataset was built as a part of the fabric defect detection project of the Intelligence Lab of University of Moratuwa, Sri Lanka. The dataset consisted of images with resolution of 1920x1080 or 1080x1920. It consisted of 398 defected images with different types of stains and 68 defect free images. Fabric Defect Dataset The dataset was also taken from Kaggle and it was supplied by the Intelligence Lab, Department of computer science and Engineering, University of Moratuwa (Ranathunga, 2020). This dataset consisted of 3 classes of defective images namely horizontal, vertical and holes along with 3 mask images for each defective image sample. The folder named 'captured' consisted of raw images. Images have a size of 640x360. Dataset from the author of the literature review paper Authors of the research papers in the literature review were contacted through emails for the dataset. With the consent of the author of the paper “Automatic fabric defect detection method using pran-net” (Peng, 2020) named Troy Peng, the annotated data was used (Peng et. al, 2020). After collecting the images from all these sources, the distribution of the images is as given below: ● Defect free 1,666 ● Defected 1,073 Defected hole 281 Defected horizontal 136 Defected lines images 157 Defected stain images 398 Defected verticle images 101 Total Images = 2,739