Installation
The EdgeLab runtime environment requires PyTorch and the following OpenMMLab third-party libraries:
MMCV: OpenMMLab Computer Vision Foundation Library
MMClassification: OpenMMLab image classification toolkit and benchmarking. In addition to classification tasks, it is also used to provide a variety of backbone networks
MMDetection: OpenMMLab detection toolbox and benchmark
MMDPose: OpenMMLab inspection toolbox and benchmark
MIM: MIM provides a unified interface for starting and installing the OpenMMLab project and its extensions, and managing the OpenMMLab model library.
Prerequisites
We strongly recommend you to use Anaconda3 to manage python packages. You can use script to configure the environment after finishing the step 0, or you can follow all the below steps to prepare the environment.
Step 0. Download and install Miniconda from the official website.
Step 1. Create a conda environment and activate it.
conda create --name edgelab python=3.8 -y
# activate edgelab
conda activate edgelab
Step 2. Install packages for GPU support and CPU support separately, it depends on your device.
On GPU platforms:
Install cuda, please refer to official install guide.
Install pytorch
# conda install conda install pytorch torchvision torchaudio pytorch-cuda=11.6 -c pytorch -c nvidia # pip install pip3 install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu116
On CPU platforms:
Install pytorch
# conda install conda install pytorch torchvision torchaudio cpuonly -c pytorch # pip install pip3 install torch torchvision torchaudio
Step 3. Install dependent libraries.
# pip install, it is not work for conda.
pip3 install -r requirements/base.txt
Step 4. Install MMCV using MIM.
pip3 install -U openmim
# must use mim install
mim install mmcv-full==1.7.0
Other method
The configuration of the project environment can be done automatically using a script on ubuntu 20.04, or manually if you are using other systems. All relevant environments can be configured on ubuntu with the following command.
python3 tools/env_config.py
Note: The above environment configuration time may vary depending on the network environment.
Reminders
After the appeal steps are completed, the required environment variables have been added to the ~/.bashrc file. A conda virtual environment named edgelab has been created and the dependencies have been installed in the virtual environment. If it is not activated at this point. You can activate conda, the virtual environment and other related environment variables with the following command.
source ~/.bashrc
conda activate edgelab