- WORKS FOR LINUX DISTRIBUTION
Docker GPU Computing Container for GIS and Remote Sensing Applications. This container is based on ubuntu-18.04 and Python 3.6
If you are not familiar with Docker, please first watch this perfect tutorial: KodeKloud Docker Tutorial (First half is enough to understand below).
In case you want to add or remove python packages go to # PART - 3 Install Python Packages
in the Dockerfile and add or remove the package below pip3 install
.
imagecodecs
jupyterlab
numpy==1.16
scipy
Pillow
matplotlib
opencv-contrib-python
scikit-image
scikit-learn
xgboost
fiona
shapely
geopandas
rasterio
tifffile
tensorflow-gpu==2.1.0
keras
- NVIDIA-docker dependencies as in here: NVIDIA-docker
In case you don't need to configure the Dockerfile
docker pull bkavlak/gisgpu:base
OR
In case you want to configure the Dockerfile
.
Go to the directory where Dockerfile
resides and type:
docker build -t 'NAME:TAG' .
sudo docker run -it --name 'CONTAINER NAME' --gpus all -p 8888:8888 -p 6006:6006 -v 'VOLUME DIRECTORY':/edenazar/data 'NAME:TAG' bash
You can add a volume to the container where you transfer files between the computer and the container.
Run command starts the container and then you can run a Jupyter Notebook as below:
jupyter notebook --ip=0.0.0.0 --port=8888 --allow-root
If successful, you will see some links below.
http://127.0.0.1:888/?token......
Copy the last link to a browser (like Chrome). Hit ENTER.
Check whether GPU is identified by the container:
import tensorflow as tf
tf.config.experimental.list_physical_devices('GPU')
Enjoy!