Data science tools and info

This is an example of a StarCraft 2 image classification. Precisely, classify the race of a unit. I am using tensorflow.js, a javascript library to load a trained model with Keras. You can see all the steps in my github repo:  https://github.com/bsaldivaremc2/keras_tutorials  The images were trained with a 64 by 64 size, so all images will be resized to this shape. Click on Choose File, then click on Classify image and see the message below. The first time it will take some seconds loading the model. Upload Image File: Classify image Empty

I haven't seen this information explicitly on the web rather than issues reported. When using a virtual environment where you have installed tensorflow-gpu and you install keras, keras will install the CPU tensorflow version. By doing this it won't be possible to use the GPU. Thus, the correct order to use keras with the GPU tensorflow version is: Install keras Install tensorflow-gpu After this, the default backend for keras will be tensorflow and will work with the GPU. To validate if you are using the GPU [1] : import tensorflow as tf sess = tf . Session ( config = tf . ConfigProto ( log_device_placement = True ))