-
-
It's interesting. I think the takeaway is "the easiest way to solve static image classification is via textures", rather than a fundamental limitation of neural nets. (evinced by, they can make neural nets do shape recognition when explicitly trying to do that)
-
Right, I think what we learned is that you can solve the image labeling problem using texture, and also that you can learn texture features using convolutions plus probably a very few layers (one?) of ANN. Whether DLs can do much with shape remains to be seen—clearly some, though
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.