Could Data Augmentation improve the performance?

[sunchengyi]

Usually in other computer vision models, the data augmentation is used in training to improve the performance on the test data.
Why is the Image augmentation not used here to train the model?
Is there any specific reason?

[skokec]

Hi, in our first journal paper we did try augmentation with rotation on KSDD1 dataset but did not find any improvements so we did not use it later.

It is possible augmentation would work on other datasets or other types of augmentation would be beneficial. In general, there is no major issue in using data augmentation. You do need to be careful which data augmentation you use in some datasets. For instance color jittering may change the color of defects in KSSD2 dataset, which could cause difficulty in recognizing specific types of defects from others.

[sunchengyi]

I see. Thank you!