Docker Environment
Features
- Three docker images: Tensorflow, Keras, Caffe
- Tensorflow object detection API
- NVIDIA GPU
- Display webcam video in docker container via opencv windows
Installation
Ubuntu 16.04
For developer environments
$ sudo ./install_docker_ce_ubuntu_amd64.sh # Install docker-ce
$ sudo ./install_nvidia_docker_ubuntu_amd64.sh # Install nvidia-docker
$ sudo mkdir /root/Data # Our convention is to put training data here. This folder will be mounted to the docker container.Create environment in Docker
Tensorflow
To build the clean Tensorflow 1.8 (GPU) environment
$ cd {PROJECT_ROOT}/framework/tensorflow/gpu
$ sudo make build # Build tensorflow DockerFileIf you want to build the clean Tensorflow 1.8 (GPU) environment
$ sudo make build TF_VER=1.8.0To run the clean Tensorflow (GPU) environment
$ cd {PROJECT_ROOT}/framework/tensorflow/gpu
$ sudo make x11 # or make x11 TF_VER=1.8.0Keras with Tensorflow backend
To build the clean Keras (GPU) environment
$ cd {PROJECT_ROOT}/framework/keras/gpu
$ sudo make build # Build keras docker fileTo run the clean Keras (GPU) environment
$ cd {PROJECT_ROOT}/framework/keras/gpu
$ sudo make x11Run the clean Keras (GPU) environment on the web (IPython notebook and TensorBoard)
$ cd {PROJECT_ROOT}/framework/keras/gpu
$ sudo make notebookYou will see the notebook and Tensorboard URL in the terminal
Caffe
To run the clean caffe environment
$ cd framework/caffe
$ sudo make bashTo Test is caffe environment correct. Let's train LeNet on MNIST
$ cd $CAFFE_ROOT
$ ./data/mnist/get_mnist.sh
$ ./examples/mnist/create_mnist.sh
$ ./examples/mnist/train_lenet.shIf you see "Optimization Done" in the console like following. Congratulation!
I1129 03:51:48.452687 39 caffe.cpp:259] Optimization Done.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent