A small program to run (but not train) a convolutional network for image classification, to accompany the blog post What does a convolutional neural net actually do when you run it?
The interesting code is in
$ make $ ./obtain-data.sh $ ./flower ./data/test/sunflowers/9783416751_b2a03920f7_n.png Classification took 0.0849622 sec sunflowers: 99.9181% dandelion: 0.0560829% tulips: 0.0258057% daisy: 1.06907e-05% roses: 2.91274e-06%
Code by Chris Cannam, copyright Queen Mary University of London. Published under MIT/X11 licence, see the file COPYING for details.