CODEWITHSUNDEEP
python-3.xtensorflowkerasconv-neural-network
Vilq
Vilq

Reputation: 601

Neural network filtering out everything else than cats/dogs

The question is simple, however I cannot find solution:

How to recognise cats and dogs and filter out everything other?

Another words: I have a big database with images of cats and dogs and all other photos mixed together, is there a way to say in output: cat -or- dog -or- something else?

Either CNNs are not constructed to do such things efficiently?

I'm using python / Keras / Tensorflow; solution with finding cats / dogs when I provide images of cats and dogs only - works fine.

Upvotes: 0

Views: 45

Answers (1)

Vilq
Vilq

Reputation: 601

Labelling next class as 'other' or something like that is the simplest way to do that.

What I found, that because of many different types of images in 'other' class, training sample should be much bigger than just when differing between cats and dogs.

Upvotes: 1

Related Questions