This is a image classification by VGG16 pre-trained model.
Python3.5
Keras2.0
TensorFlow
Windows10
Most py-package (such like numpy or cv2...) you can install by pip install xx.whl from here.
Download the weight file 'vgg16_weights_tf_dim_ordering_tf_kernels.h5'. In here and download it.
Put the weight file into data/model
model/synset_words.txt : map class id (which is equal to line id) to synset_id and human words. For example, If you get the prediction id is 281, then you should out put the 281th content in synset_words.txt. model/vgg16_weights_tf_dim_ordering_tf_kernels.h5: special VGG16 pre-trained weight file for Keras2.
Mind the path which is used in windows. If you are UNIX player, should change the format.
Any other errors happen you can check my blog.
You should be careful with the Keras's backend (th:theano or tf:tensorflow), because they have different mat size format.
About model you can find in Github-Glist.
About VGG16 network structure find in here.
Give me a star. Thks.