Getting started with deep learning, many people have experienced the bitter journey from beginner to abandonment, with countless individuals failing at the first hurdle: environment configuration issues. As the saying goes, if the environment is not set up correctly, learning leads to tears.

If you are facing the pain of configuring the environment, whether you are a Windows user, Ubuntu user, or a die-hard Mac fan, this article is tailor-made for you. Next, we will discuss the deep learning environment configuration issues for Windows, Mac, and Ubuntu in turn.

1. Windows System Deep Learning Environment Configuration

Installation Combination: Anaconda + PyTorch (GPU version) + GTX1060

1.1 Open Anaconda Prompt

1.2 Determine the CUDA version supported by hardware

NVIDIA Control Panel – Help – System Information – Components

1.3 Determine PyTorch version and torchvision version

Visit the PyTorch official website: https://pytorch.org/get-started/locally/

Because the official source is too slow, here we use Tsinghua source for downloading

1.4 Download the corresponding installation package from the mirror

Tsinghua Mirror:

PyTorch:

torchvision:

1.5 Local Installation

Then return to the directory where the virtual environment is located and use conda install anaconda to install the basic packages required for the environment

1.6 Test

Code 1:

from __future__ import print_function
import torch
x = torch.rand(5, 3)
print(x)

import torch
torch.cuda.is_available()

If there are no exceptions in the output of the above two code snippets, it indicates that the environment setup was successful.

1.7 Issues Encountered

• Error creating virtual environment with the command below

conda create -n py37_torch131 python=3.7

The environment will be saved in the envs folder under the Anaconda directory

• PackagesNotFoundError: The following packages are not available from current channels

• Conda download is too slow

I shout Tsinghua is awesome!

• CUDA installation

• PyTorch installation

conda install pytorch torchvision cudatoolkit=10.0 -c pytorch

• How to enter the virtual environment in Jupyter Notebook

Delete kernel

python -m ipykernel install --user --name py27-caffe-notebook

Thus, the kernel addition is complete. To view existing kernels:

jupyter kernelspec list

Delete an existing kernel

jupyter kernelspec remove <kernel_name></kernel_name>

The above command only deletes the configuration file and does not uninstall the corresponding ipykernel of the virtual environment. Therefore, if you want to reinstall the kernel for the corresponding Python virtual environment, just activate the virtual environment, and then

python -m ipykernel install --name <kernel_name></kernel_name>

• Conda installation fails halfway

Download the file locally, navigate to the directory of the file, and use the command

conda install --offline <package_name></package_name>

2.1 Install Anaconda

2.2 Confirm Download Status

In the Mac Terminal, enter

python --version

Ensure that the installed Python is version 3.x. In the Terminal, enter

jupyter notebook

2.2.1 Common Issues

If multiple Anaconda versions are downloaded on the computer, conflicts may occur during operation. In the Terminal, enter

cd ~

Return to the home directory and enter

cat .bash_profile

If you can only see one version of Anaconda, there is no problem. If there are multiple versions, conflicts may occur when downloading packages. Use vim, nano, or other text editors to delete the old version of Anaconda’s

export PATH= ...

Delete it.

2.3 Creating Virtual Environments and Downloading Packages

Using conda to create virtual environments and download corresponding packages is a simple task.

2.3.1 Graphic

Then on the Home page, ensure that the top left corner points to the environment name you just created. In this environment, install jupyter notebook, noting that the original notebook installed is in base and cannot be shared.

Return to the Environments section, and you can see all the packages in this environment. Select All in the upper left corner, and enter the name of the package you want to download

Then select it for download

2.3.2 Command Line

Using the command line to complete the above operations is also very straightforward. This time we take PyTorch as an example. In the Terminal, enter

conda create --name env_name

This will create a virtual environment called “env_name”. Enter

conda env list

to see all created virtual environments, where the one with * is the current environment. Enter

conda activate env_name

to enter the environment

conda deactivate

to exit the current virtual environment and return to base.

2.3.3 Download PyTorch

conda install pytorch torchvision -c pytorch

Copy it, paste it into the Terminal, and run it to start downloading.

After downloading, enter in the Terminal

python3

Then import the two packages just downloaded to confirm completion

import torch
import torchvision
print(torch.__version__)
print(torchvision.__version__)

quit()

And it’s done.

2.4 Additional Situations

Click on the 64-Bit (x86) Installer (522 MB) below to download the 64-bit version.

The downloaded file name is: Anaconda3-2020.02-Linux-x86_64.sh.

cd to the directory where Anaconda3-2020.02-Linux-x86_64.sh is located:

Next, manually add the environment variable. First, cd to ~, then edit the .bashrc file: sudo vim .bashrc

# Distinguish between anaconda python and system built-in python
alias python3="/usr/bin/python3.5"
alias python2="/usr/bin/python2.7"
. /home/cxd/anaconda3/etc/profile.d/conda.sh

Enter source .bashrc to execute the modified initialization document

3.2 PyTorch CPU Version Installation

Open the PyTorch official website: https://pytorch.org/

conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/pytorch

conda install pytorch torchvision cpuonly -c pytorch

import torch
print(torch.__version__)

3.3 PyTorch GPU Installation

3.3.1 GPU Driver Installation

• Check Graphics Card Type

Execute the command: ‘ ubuntu-drivers devices’

== /sys/devices/pci0000:00/0000:00:01.0/0000:01:00.0 ==
modalias : pci:v000010DEd00001C8Dsv00001028sd0000086Fbc03sc02i00
vendor   : NVIDIA Corporation
model    : GP107M [GeForce GTX 1050 Mobile]
driver   : nvidia-driver-390 - distro non-free
driver   : nvidia-driver-435 - distro non-free
driver   : nvidia-driver-440 - distro non-free recommended
driver   : xserver-xorg-video-nouveau - distro free builtin

== /sys/devices/pci0000:00/0000:00:14.3 ==
modalias : pci:v00008086d0000A370sv00008086sd000042A4bc02sc80i00
vendor   : Intel Corporation
model    : Wireless-AC 9560 [Jefferson Peak]
manual_install: True
driver   : backport-iwlwifi-dkms - distro free

You can see here that there is a device which is GTX 1050. The recommended driver is 440.

• Install Driver

Install all recommended drivers

sudo ubuntu-drivers autoinstall

To install one driver

sudo apt install nvidia-440

3.3.2 Install CUDA

CUDA installation requires the appropriate graphics card driver. Below is the version correspondence between the driver and CUDA

Table 1. CUDA Toolkit and Compatible Driver Versions
CUDA Toolkit    Linux x86_64 Driver Version Windows x86_64 Driver Version
CUDA 10.2.89    >= 440.33   >= 441.22
CUDA 10.1 (10.1.105 general release, and updates)   >= 418.39   >= 418.96
CUDA 10.0.130   >= 410.48   >= 411.31
CUDA 9.2 (9.2.148 Update 1) >= 396.37   >= 398.26
CUDA 9.2 (9.2.88)   >= 396.26   >= 397.44
CUDA 9.1 (9.1.85)   >= 390.46   >= 391.29
CUDA 9.0 (9.0.76)   >= 384.81   >= 385.54
CUDA 8.0 (8.0.61 GA2)   >= 375.26   >= 376.51
CUDA 8.0 (8.0.44)   >= 367.48   >= 369.30
CUDA 7.5 (7.5.16)   >= 352.31   >= 353.66
CUDA 7.0 (7.0.28)   >= 346.46   >= 347.62

CUDA download link: http://suo.im/6dY8rL

Installer Type can choose either of the first two. However, after obtaining the CUDA file, first check the gcc version. Below is an example using the first runfile (local) installation method.

• Install GCC

Linux generally comes with GCC, let’s first check the version of GCC in your system

gcc --version

And the GCC dependency version for CUDA will be provided in the official installation guide

sudo apt-get install gcc-7.0
sudo apt-get install g++-7.0

After installation, you need to switch the system GCC version

sudo update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-7 50
sudo update-alternatives --install /usr/bin/g++ g++ /usr/bin/g++-7 50

Select the required version

sudo update-alternatives --config gcc
  Select       Path          Priority  Status
------------------------------------------------------------* 0            /usr/bin/gcc-9   50        Automatic mode  1            /usr/bin/g++-9   50        Manual mode  2            /usr/bin/gcc-7   50        Manual mode

• Install CUDA

sudo sh cuda_<your_version>_linux.run</your_version>

• Configure Environment Variables

sudo vim ~/.bashrc

Copy the following command into it

export PATH=/usr/local/cuda-10.2/bin${PATH:+:$PATH}}
export LD_LIBRARY_PATH=/usr/local/cuda-10.2/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}

Check if the installation was successful

nvcc -V

3.3.3 Install cuDNN

https://developer.nvidia.com/cudnn Select the version corresponding to CUDA

Then copy the include and lib64 folders of cuDNN to CUDA

sudo cp cuda/include/cudnn.h    /usr/local/cuda-10.2/include    # The name of the unzipped folder is cuda-10.2
sudo cp cuda/lib64/libcudnn*    /usr/local/cuda-10.2/lib64
sudo chmod a+r /usr/local/cuda-10.2/include/cudnn.h   /usr/local/cuda-10.2/lib64/libcudnn*

3.3.4 Install PyTorch-GPU

Conda installation:

# Choose the corresponding CUDA version
conda install pytorch torchvision cudatoolkit=10.2

pip installation:

pip install torch torchvision -i  https://pypi.mirrors.ustc.edu.cn/simple

Download 1: OpenCV-Contrib Extension Module Chinese Version Tutorial
Reply "Extension Module Chinese Tutorial" in the "Beginner Learning Visual" public account backstage to download the first OpenCV extension module tutorial in Chinese, covering more than twenty chapters including extension module installation, SFM algorithms, stereo vision, target tracking, biological vision, super-resolution processing, etc.

Download 2: Python Vision Practical Projects 52 Lectures
Reply "Python Vision Practical Projects" in the "Beginner Learning Visual" public account backstage to download 31 visual practical projects including image segmentation, mask detection, lane line detection, vehicle counting, eyeliner addition, license plate recognition, character recognition, emotion detection, text content extraction, face recognition, etc., to help you quickly learn computer vision.

Download 3: OpenCV Practical Projects 20 Lectures
Reply "OpenCV Practical Projects 20 Lectures" in the "Beginner Learning Visual" public account backstage to download 20 practical projects based on OpenCV, achieving advanced learning in OpenCV.

Communication Group

Welcome to join the public account reader group to communicate with peers. Currently, there are WeChat groups for SLAM, 3D vision, sensors, autonomous driving, computational photography, detection, segmentation, recognition, medical imaging, GAN, algorithm competitions, etc. (will gradually be subdivided). Please scan the WeChat ID below to join the group, and note: "Nickname + School/Company + Research Direction", for example: "Zhang San + Shanghai Jiao Tong University + Visual SLAM". Please follow the format; otherwise, the request will not be approved. After successfully adding, you will be invited into the relevant WeChat group based on your research direction. Please do not send advertisements in the group; otherwise, you will be removed from the group. Thank you for your understanding~