Convolutional Neural Networks with Caffe and NEGATIVE IMAGES

0
votes

When training a set of classes (let's say #clases (number of classes) = N) on Caffe Deep Learning (or any CNN framework) and I make a query to the caffemodel, I get a % of probability of that image could be OK.

So, let's take a picture of a similar Class 1, and I get the result:

1.- 90%

2.- 10%

rest... 0%

the problem is: when I take a random picture (for example of my environment), I keep getting the same result, where one of the class is predominant (>90% probability) but it doesn't belong to any class.

So what I'd like to hear is opinions/answers from people which has experienced this and would have solved how to deal with no-sense inputs to the Neural Network.

My purposes are:

  1. Train one more extra class with negative images (like with train_cascade).
  2. Train one more class extra with all the positive images in the TRAIN set, and the negative on the VAL set.

But my purposes don't have any scientific base to execute them, that's why I ask you this question.

What would you do?

Thank you very much in advance.

Rafael.

machine-learningneural-networkdeep-learningcaffeconv-neural-network
have you sorted this out yet? - user1269942
No... still looking for it. But if you train many classes and they are ok, when you pass a non-result you get a balance result for each class, so you could deal with it in that way maybe - Rafael Ruiz Muñoz
I recently trained a deep nn to do something similar...presence/absence detection. I just made as many absent training images as I had present images. It is a bit of an ask for a network to 'detect' nothing, however...but eventually the results were good. For a multi-class situation, I would try just adding a 'nothing' class and get some data to train with. - user1269942
have you try to pass that image to the ImageNet ??? what's your definition of "nothing" ? - Rafael Ruiz Muñoz
for me, "nothing" is an image without an animal in it. "something" is an image with an animal. - user1269942