0
$\begingroup$

I have to do a project that detects fabric surface errors and I will use machine learning methods to deal with it. I have a dataset that includes around six thousand fabric surface images with the size 256x256. This dataset is labeled, one thousand of it was labeled as NOK that means fabric surface with error, and the rest was labeled as OK which means fabric surface without an error.
I read a lot of papers about fabric surface error detection with machine learning methods, and I saw that "autoencoders" are used to do it. But as I saw that the autoencoders are used in unsupervised learning models without labels. I need to do it with supervised learning models. Is there any model that can I use for fabric surface error detection with images in the supervised learning? Can be autoencoders used for it or is there any better model to do it?

$\endgroup$
2
  • $\begingroup$ Have you already thought about Convolutional NNs? (CNNs) It sounds like you have a reasonable amount of data, so you even could start from scratch with a custom built CNN. If you are lucky maybe you can find a pretrained model for a similar problem, which you can build on. $\endgroup$
    – jottbe
    Dec 1, 2020 at 20:47
  • $\begingroup$ Thanks a lot. I am working on the CNNs models right now. I think it can be useful for this project. $\endgroup$ Dec 4, 2020 at 19:27

2 Answers 2

1
$\begingroup$

Papers With Code have a great summary of the tasks in computer vision and their respective State of the Art models. Also the tasks page from HuggingFace serves as a great reference.

$\endgroup$
0
$\begingroup$

Convolutional Neural Networks are mostly used for all kind of computer vision tasks.

Here you can find a tutorial on how to train a CNN for image classification from scratch.

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .