Deconvnet

The is a fork of jalused's implementation restructured and updated to be compatible with Keras 2.0 and a tensorflow backend.

It implements a Deconvnet in keras, following Matthew D.Zeiler's paper Visualizing and Understanding Convolutional Networks

Feature

Given a pre-trained keras model, this repo can visualize features of specified layer including dense layer.

Dependencies

• Keras >= 2.1 (Tensorflow Backend)
• Python >= 3.4

Shortage

• The code implements visualize function for only Convolution2D, MaxPooling2D, Flatten, Input, Activation layers, thus cannot handle other type of layers.
• The code support only plain networks, thus cannot visualize ResNet, Highway Networks or something.