most people/businesses interested in image classification don't have large datasets, yet there's little material about small data
Today I'm writing a post on how to learn to classify cat and dog pictures given only 1000 examples of each. Diff. approaches, 80-95% acc.
-
-
-
have you seen a fully nonparametric Bayesian convnet-like solution?
- Show replies
New conversation -
-
-
Looking forward to this. This will also lower the bar on teaching people deep learning.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
hehe "only" :) sounds cool though. how many do you need to get 75% on non-strange-looking-examples?
-
this is from a Kaggle dataset that had 25k images, so it's small in comparison. It's also collectable by hand in an afternoon...
End of conversation
New conversation -
-
-
can you post a link please?
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
looking forward to reading it
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Eager to read about small datasets with Keras ! Plz post the link when it's done...
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
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.