Freezing models in Tensorflow
Not my original post. However made some updates minor in the posts with latest version Freeze Tensorflow models and serve on web In this tutorial, we shall learn how to freeze a trained Tensorflow Model and serve it on a webserver. You can do this for any network you have trained but we shall use the trained model for dog/cat classification in this earlier tutorial and serve it on a python Flask webserver. So you trained a new model using Tensorflow and now you want to show it off to your friends and colleagues on a website for a hackathon or a new startup idea. Let’s see how to achieve that. 1. What is Freezing Tensorflow models? Neural networks are computationally very expensive. Let’s look at the architecture of alexnet which is relatively simple neural network: Let’s calculate the number of variables required for prediction: conv1 layer: ( 11*11)*3*96 (weights) + 96 (biases) ...
