An interest of mine is anomaly detection in images.
I am currently experimenting with fully connected auto encoders (unsupervised) but I wondered if the fastai library (supervised) presented another way into this subject.
Typically the training set will comprise of images of good product, with the expectation that the trained model will discriminate against images of containing bad product artefacts and heat map the offending section.
Any direction would be appreciated. Thanks in advance.
As this is an interest of mine also, I’m glad you brought it up. Perhaps you’ve already seen it, but i’ll refer you to the thread from 2017 on this subject where I shared my experience tackling this problem.
A nice sanity check that we added was utilizing instance-based learning with histograms of the images, which was relevant for my use-case (images from pretty much the same angle of the same object)