I've got classification problem on image, I have 10 classes and when I fine tuned my model on it (I tried VGG, Xception, resnet etc) I have approximatly 83% validation accuracy.

I was wondering if doing lot of binary model with 1 class represented and the other as 'other' and then using them to classify my image would be good and efficient ? (I obtain more than 90% val acc for each class doing this)

Except for memory consumption and training time does this method have drawback ?

  • $\begingroup$ Well it's inefficient use of your CNN weights for sure, earlier you were using it to draw multiple decision boundaries now you draw only 1, which implies over fitting might be a problem. $\endgroup$ – DuttaA Mar 18 '19 at 16:21

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.