Skip to content

This is a dataset of our paper "A compact convolutional neural network for surface defect inspection"

Notifications You must be signed in to change notification settings

abin24/Textures-Dataset

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

9 Commits
 
 
 
 
 
 

Repository files navigation

Textures Classification dataset

We integrate a dataset of surface texture (8674 images), which contains 64 classes from 3 public datasets. The images are uniformly resized into 331 x 331. Notably, 11 classes (3194 images) from KTH surface datast ( including KTH-TIPS and KTH-TIPS2) , 28 classes (4480 without rotated patches) from Kyberge dataset, and 25 classes(1000 images) from the UIUC dataset dataset. Details are displayed in figure.

These images are real-world surface texture from wood, blanket, cloth, leather, etc, it can be used to evaluate the capacity of the models or used as a pretrain dataset to improve the performance of the CNN models for surface defect inspection.

The textures dataset. Each image is a sample of one texture class, first row is from the KTH, second row from Kyberge, and third from UIUC.

the dataset is split into half randomly, one for the training set, and another for the validation. It is used in our upcoming paper "A compact convolutional neural network for surface defect inspection".

you can download the dataset from baiduyun with Extraction code 513a

or Gogle Drive

If you have any difficult in access these datasets, contact me by [email protected] or [email protected]

About

This is a dataset of our paper "A compact convolutional neural network for surface defect inspection"

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published