News for hackers. Stuff that matters.
Re:Where we see shapes, AI sees textures by udayrddy on Monday July 08, 2019 @01:07PM (#20384492)
So, a person with a gray texture of elephant skin shirt is classified as elephant ?!
Re:Where we see shapes, AI sees textures by Sharlin on Monday July 08, 2019 @01:57PM (#20385079)
One of the most archetypal failure modes of current NN-based classifiers is, for example, happily reporting seeing a spotted feline given a picture of a couch with a leopard pattern.
Re:Where we see shapes, AI sees textures by pmontra on Tuesday July 09, 2019 @01:05AM (#20389155)
Well, if I'm given the task to enumerate what I see in that image maybe I would also add the feline label. But I would add the picture_of attribute, if that's a thing.
Re:Where we see shapes, AI sees textures by Sharlin on Tuesday July 09, 2019 @03:01AM (#20389572)
I don’t think today’s classifiers are anywhere sophisticated enough to have a concept of something vs. a picture of something, a type of use–mention distinction, or ceci n’est un pipe if you will. The point was that people think they trained their classifier to recognize leopards, but in actuality it just learned to recognize the leopard coat pattern.
Re:Where we see shapes, AI sees textures by liuliu on Monday July 08, 2019 @01:27PM (#20384734)
Yes, maybe. But it is at least unclear whether is inherently CNN or not (many people would vote for no). ImageNet training data is heavily texture dependent, therefore, the trained model leaning towards texture information more. Some recent papers did style transfer to randomize textures in order to train the model to focus more on the shape, the reported results seem good.
Re:Where we see shapes, AI sees textures by w_t_payne on Monday July 08, 2019 @01:35PM (#20384818)
Exactly. I would also add (although without strong evidence to support my assertion) that it seems to be more a function of the training data than of the network architecture.