TensorFlow series: How to use inception ResNet v2 Network

First, the foreword recently in the Inception V3 and Inception ResNet v2 These two networks, these two network architectures I don't think I said more, Google produced. By fusing the feature map of different scales to replace the nxn convolution by 1xn convolution kernel nx1 convolution, the computational volume is effectively reduced, and the computational volume is reduced by using multiple 3x3 convolution instead of 5x5 convolution and 7x7 convolution. In addition, the network structure of ResNet and inception is fused in Inception ResNet v2 to further promote imagenet on accuracy.
Nonsense so much, since to actually use these two networks we certainly want to use Google's public code directly to Finetune, so that not only can save time, but also to get good results, here you need to use the latest Slim library to implement the inception ResNet v2 network.
Of course, the latest tensorflow (1.2.0) does not have the structure of the network, its own only inception V3 the network structure.

Second, how to use the latest Slim library Reference:https://github.com/tensorflow/models/tree/master/slim#install

(1) Download code (please own proxy server, use proxy server download will be much faster) CD ~/projects git clone https://github.com/tensorflow/models/
(2) Compile the standalone Slim package and install the CD Models/slim python setup.py Build
sudo python setup.py install
(3) How to use such as using the INCEPTION_RESNET_V2 network can be so imported from nets import Inception_resesnet_v2
Of course, the specific usage can refer to reference best to flip the code of the Slim library. Easier to understand here is an example to refer to Https://github.com/kwotsin/transfer_learning_tutorial