Stack of GPU Docker files based on nvidia/cuda:9.0-cudnn7-devel-ubuntu16.04. All of the docker images has been tested on Azure GPU VM NC6

If you have nvidia-docker 1.0 installed: we need to remove it and all existing GPU containers

$ sudo docker volume ls -q -f driver=nvidia-docker | xargs -r -I{} -n1 docker ps -q -a -f volume={} | xargs -r docker rm -f

$ sudo apt-get purge -y nvidia-docker

Add the package repositories

curl -s -L https://nvidia.github.io/nvidia-docker/gpgkey | sudo apt-key add -
distribution=$(. /etc/os-release;echo $ID$VERSION_ID)
curl -s -L https://nvidia.github.io/nvidia-docker/$distribution/nvidia-docker.list | sudo tee /etc/apt/sources.list.d/nvidia-docker.list

$ sudo apt-get update

Install nvidia-docker2 and reload the Docker daemon configuration

$ sudo apt-get install -y nvidia-docker2

$ sudo pkill -SIGHUP dockerd

Test nvidia-smi with the latest official CUDA image

$ sudo docker run --runtime=nvidia --rm nvidia/cuda:9.0-base nvidia-smi

$ sudo docker login

#test jupyter $ sudo docker pull jupyter/minimal-notebook

$ sudo docker run --rm -p 8888:8888 -v "$PWD":/home/jovyan/work jupyter/minimal-notebook

#set up your workspace folder - ideally on the attached disk $ mkdir myworkspace

$ cd myworkspace/

$ chmod 777 $PWD

pull image from docker hub

basic gpu jupyter notebook

$ sudo docker pull azadehkhojandi/gpu-minimal-notebook

$ sudo nvidia-docker run --rm -p 8888:8888 -v "$PWD":/home/jovyan/work azadehkhojandi/gpu-minimal-notebook

mask rcnn pytorch notebook

$ sudo docker pull azadehkhojandi/gpu-maskrcnn-pytorch-notebook

$ sudo nvidia-docker run --rm -p 8888:8888 -v "$PWD":/home/jovyan/work azadehkhojandi/gpu-maskrcnn-pytorch-notebook

GPU VM - Azure NC6

add port 8888 into network panel http://{publicipofvm}:8888/?token={token after running azadehkhojandi/pytorchgpujupyter } check gpu on vm $ lsb_release -a

Check Pytorch and cuda - inside container

$ nvcc --version

import torch
print('Torch version:',torch.__version__)
print('Is cuda available:',torch.cuda.is_available())
if torch.cuda.is_available():
    print('Graphic card name:',torch.cuda.get_device_name(torch.cuda.current_device()))

If torch can not be found run pip install torch

Mask RCNN - inside container

From the folder myworkspace created above: $ git clone https://github.com/multimodallearning/pytorch-mask-rcnn.git

for NC6 | Tesla K80 | sm_37 |

cd pytorch-mask-rcnn
cd nms/src/cuda/
nvcc -c -o nms_kernel.cu.o nms_kernel.cu -x cu -Xcompiler -fPIC -arch=sm_37
cd ../../
python build.py
cd ../

cd roialign/roi_align/src/cuda/
nvcc -c -o crop_and_resize_kernel.cu.o crop_and_resize_kernel.cu -x cu -Xcompiler -fPIC -arch=sm_37
cd ../../
python build.py
cd ../../

When running build.py, if you received the error:

/home/jovyan/work/myworkspace/pytorch-mask-rcnn/roialign/roi_align/src/crop_and_resize.c:124:33: error: dereferencing pointer to incomplete type ‘THTensor {aka struct THTensor}’
    const int batch_size = image->size[0];

Run the command pip install http://download.pytorch.org/whl/cu91/torch-0.4.0-cp36-cp36m-linux_x86_64.whl rather than the standard pip install torch The github issues found here: longcw/RoIAlign.pytorch#11 say it can be fixed by installing pytorch

$ git clone https://github.com/cocodataset/cocoapi.git

$ cd cocoapi/PythonAPI
make
cd ../..

$ wget 'https://azpublicblob.blob.core.windows.net/public/mask_rcnn_coco.pth'

$ ln -s cocoapi/PythonAPI/pycocotools pycocotools

$ pip install scikit-image matplotlib scipy h5py

$ python demo.py

Install Az copy - inside container

https://docs.microsoft.com/en-us/azure/storage/common/storage-use-azcopy-linux In order to do the following steps, the docker container must be launched with sudo nvidia-docker run --rm --privileged -p 8888:8888 -e GRANT_SUDO=yes --user root -v "$PWD":/home/jovyan/work azadehkhojandi/gpu-minimal-notebook. This grants sudo privileges inside of the container

$ echo "deb [arch=amd64] https://packages.microsoft.com/repos/microsoft-ubuntu-xenial-prod/ xenial main" > azure.list

$ sudo cp ./azure.list /etc/apt/sources.list.d/

$ sudo apt-key adv --keyserver packages.microsoft.com --recv-keys EB3E94ADBE1229CF

$ sudo apt-get update

$ sudo apt-get install azcopy

$ nano demo.py

update "# Visualize results" to following

# Visualize results
r = results[0]
print(r['rois'])
print(r['masks'])
print(r['class_ids'])
print(class_names)
print(r['scores'])
visualize.display_instances(image, r['rois'], r['masks'], r['class_ids'],
                            class_names, r['scores'])
plt.savefig('result.jpg')

python demo.py you should be able to see 'result.jpg' created

cv2 pip install opencv-python

pip install azure pip install azure-storage --upgrade

• note => replace azadehkhojandi with your dockerhub or azure container regisetry username

Buidling gpu-minimal-notebook image from docker file

$ sudo nvidia-docker build -t azadehkhojandi/gpu-minimal-notebook -f gpu-minimal-notebook.dockerfile .

$ sudo docker image list

$ sudo nvidia-docker tag {imageid} azadehkhojandi/gpu-minimal-notebook:barebone

$ sudo nvidia-docker push azadehkhojandi/gpu-minimal-notebook

$ sudo nvidia-docker run --rm -p 8888:8888 -v "$PWD":/home/jovyan/work azadehkhojandi/gpu-minimal-notebook

Buidling gpu-pytorch-notebook image from docker file

$ sudo nvidia-docker build -t azadehkhojandi/gpu-pytorch-notebook -f gpu-pytorch-notebook.dockerfile .

$ sudo docker image list

$ sudo nvidia-docker tag {imageid} azadehkhojandi/gpu-pytorch-notebook:barebone

$ sudo nvidia-docker push azadehkhojandi/gpu-pytorch-notebook

$ sudo nvidia-docker run --rm -p 8888:8888 -v "$PWD":/home/jovyan/work azadehkhojandi/gpu-pytorch-notebook

Buidling pu-objectdetection-tensorflow-notebook image from docker file

• wip - unstable $ sudo nvidia-docker build -t azadehkhojandi/gpu-objectdetection-tensorflow-notebook -f gpu-objectdetection-tensorflow-notebook.dockerfile .

$ sudo docker image list

$ sudo nvidia-docker tag {imageid} azadehkhojandi/gpu-objectdetection-tensorflow-notebook:barebone

$ sudo nvidia-docker push azadehkhojandi/gpu-objectdetection-tensorflow-notebook

$ sudo nvidia-docker run --rm -p 8888:8888 -v "$PWD":/home/jovyan/work azadehkhojandi/gpu-objectdetection-tensorflow-notebook

Buidling gpu-maskrcnn-pytorch-notebook image from docker file

• wip - unstable $ sudo nvidia-docker build -t azadehkhojandi/gpu-maskrcnn-pytorch-notebook -f gpu-maskrcnn-pytorch-notebook.dockerfile .

$ sudo docker image list

$ sudo nvidia-docker tag {imageid} azadehkhojandi/gpu-maskrcnn-pytorch-notebook:barebone

$ sudo nvidia-docker push azadehkhojandi/gpu-maskrcnn-pytorch-notebook

$ sudo nvidia-docker run --rm -p 8888:8888 -v "$PWD":/home/jovyan/work azadehkhojandi/gpu-maskrcnn-pytorch-notebook

References

https://azure.microsoft.com/en-us/global-infrastructure/services/?products=virtual-machines

https://github.com/NVIDIA/nvidia-docker

http://arnon.dk/matching-sm-architectures-arch-and-gencode-for-various-nvidia-cards/

https://github.com/jupyter/docker-stacks/blob/master/base-notebook/Dockerfile

https://github.com/jupyter/docker-stacks/tree/master/minimal-notebook

http://goinbigdata.com/docker-run-vs-cmd-vs-entrypoint/

https://towardsdatascience.com/tensorflow-object-detection-with-docker-from-scratch-5e015b639b0b