I have a big set of images (>10.000), where there are similarities among them. I need to find a number/group of image patterns (eg, 5) that represent all images.

As I do not know what patterns are, therefore I suppose that I need an unsupervised neural network. However, as I do not know how many patterns are, I cannot specify the number of classes/labels in this network.

Reading, I think an Autoencoder network may be useful because this network is unsupervised and I don't need to specify the number of classes. However, I am not sure if this network can infer patterns (as abstractions from images) from a dataset.

BTW, I am using Tensorflow & Keras in Python.

  • $\begingroup$ stats.stackexchange.com/questions/407267/… could be useful to you $\endgroup$ – Jan Kukacka Jul 26 at 11:23
  • $\begingroup$ Use a dimensionality reduction algorithm (Autoencoder, PCA, etc.), then apply clustering on the latent vectors. The numbers of clusters will represent the group/categories of the images. $\endgroup$ – Bloc97 Jul 31 at 21:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.