使用 Python、Tensorflow 进行图像分类

Question

How to perform Image classification. Get Breakhis data set and classify malignant vs benign using deep learning.
I have to use just Python and Tensorflow. I need to write code for data set splitting, for preprocessing and training, and for creating a prediction on a test image. How to start and how to code?

Answer 1

I was pretty much in your same condition like a week ago, so here i go.

1. Prepare some coffee
2. You might want to get everything running and ready to work first. Start by having a good python 3.x or 2.7 enviroment, I personally recommend Anaconda since it has lots of useful packages you might need. After that you want to have Tensorflow up and running, have a look at Tensorflow Instalation guide , but don't install it yet, just the requirements like CUDA if you're willing to use tensorflow with the processing power of a Nvidia GPU.

3. When you have all set to install tensorflow, just create a virtual enviroment where you like the most with

   conda create -n <myenv> python=xx

where xx is the python version you are willing to use, and is the name of your enviroment (Take it as your workshop, with all the tools you need, and where you can make a mess without damaging your computer). After that you can install Tensorflow and maybe Keras (a usefull python module that works on neural networks with Tensorflow as a backend) with pip install --upgrade tensorflow-gpu and pip install keras . or maybe use pip3 if you're willing to use python 3.x.

After this all you have to do is read and read, start from the basics, look forward to create your first neural network and so on.

Answer 2

You can visit this blog to understand basic Image Classification using TensorFlow in Python:

https://medium.com/data-science-everywhere/mnist-image-classification-using-tensorflow-29c5e48a747d