I obtained a dataset myself and performed transfer learning on the ResNet-18 network, and the results were accurate. However, I found out that I need to classify objects that don’t belong to the classifiable classes which I will name it an unknown class.
I’ve searched the internet for solutions, and but no solution looked possible for me. I thought about training random photos under the unknown label, but I don’t think that way is effective.
Please tell me if there is a way to classify unknown objects.
Thanks!
Henry