Installing bleeding edge open source software on a windows machine can end up being very challenging. In this gist I will list out the steps needed to install Keras and Tensorflow in windows machine.
- A Windows 10 machine
- Anaconda python distribution
- Open anaconda prompt on your computer (you can search for "anaconda prompt" in the search bar)
- Type the following in the anaconda prompt and hot enter:
conda create --name tensorflow python=3.5
- The previous step will create a new python environment, with python 3.5, named as "tensorflow" (you can name this environment to any name that you deem fit, tensorflow is just a name here)
- Now activate your newly created python environment by typing and hitting enter:
activate tensorflow
- Now we will install the relevant libraries and packages in this environment
- First we will install jupyter by typing the following into the console:
conda install jupyter
- Then we will install scipy:
conda install scipy
- After this we will install pandas
conda install pandas
- Then we will install seaborn
conda install seaborn
- Then, scikitlearn
conda install scikit-learn
- Then spyder (optional)
conda install spyder
- Now tensorflow
pip install tensorflow
- Finally keras
conda install -c conda-forge keras
Now run type:
ipython
This will open up an ipython prompt, within this prompt type:
import tensorflow as tf
import keras
If the above runs without error, then you have successfully installed tensorflow and keras. Now, to run tensorflow or keras, all you need to do is go to anaconda prompt and type:
activate tensorflow
jupyter notebook
This will start a jupyter notebook in which you can run tensorflow code.
In case there are any issues with the process described here, please leave a comment. I will be actively maintaining this gist,
hi
I will be very thankful if you will help me out with this..
can we convert folder which is having images into csv file or dataframe..
and how we can do that ..
the images in the folder are in png .format
Thanks
Regards
Jyotsana