or: Approximating CNNs With Bag-Of-Local-Features Models Works Surprisingly Well On Imagenet

A bag-of-deep-texture-features model works about as well as a deep ResNet on image classification. Looks like they’re not really merging global shape information..