Skip to content

Latest commit

 

History

History
161 lines (111 loc) · 7.02 KB

installation.md

File metadata and controls

161 lines (111 loc) · 7.02 KB
Raw

Installation

Language: 简体中文 | English

Prerequisites

  • GPU
  • Linux platform (Ubuntu and CentOS are recommended, Ubuntu 16.04, Ubuntu 18.04 and CentOS 7.6 have been tested.)
  • Anaconda3
  • Python 3.6+ (Python 3.7 is recommended and has been tested.)
  • PyTorch 1.3+ (PyTorch 1.6 is recommended and has been tested.)
  • CUDA 9.2+ (CUDA 10.2 is recommended and has been tested. If you build PyTorch from source, CUDA 9.0 is also compatible.)
  • CuDNN (Optional. CuDNN 7.6.5 has been tested.)
  • GCC 5+ (GCC 4.8.5, 5.5.0 and 7.5.0 are recommended and have been tested.)
  • MMCV (MMCV 1.0.5 is highly recommended and have been tested. It is the only MMCV version compatible with MMDetection 2.3.0, which is the base of this MI-AOD repository.)

Here are the compatible MMDetection and MMCV versions. Please install the correct version of MMCV to avoid installation issues.

Note: You need to run pip uninstall mmcv first if you have mmcv installed. If mmcv and mmcv-full are both installed, there will be ModuleNotFoundError.

Environment Installation

Before everything starts, please ensure that you already have Anaconda3 and CUDA 10.2 installed.

  1. Please create a conda virtual environment with Python 3.7 and activate it.

    conda create -n miaod python=3.7 -y
conda activate miaod

  2. Please install PyTorch 1.6.0 and torchvision 0.7.0 for CUDA 10.2 following the official instructions, e.g.,

    conda install pytorch==1.6.0 torchvision==0.7.0 cudatoolkit=10.2 -c pytorch

    Please note: make sure that your compilation CUDA version and runtime CUDA version match. You can check the supported CUDA version for precompiled packages on the PyTorch website.

    If you build PyTorch from source instead of installing the prebuilt package, you can use more CUDA versions such as 9.0.

  3. Please install mmcv-full, we recommend you to install the pre-build package as below.

    pip install mmcv-full=={mmcv_version} -f https://download.openmmlab.com/mmcv/dist/{cu_version}/{torch_version}/index.html

    Please replace {mmcv_version}, {cu_version} and {torch_version} in the url to your desired one. Our recommended is, to install the mmcv-full 1.0.5 with CUDA 10.2 and PyTorch 1.6.0, use the following command:

    pip install mmcv-full==1.0.5 -f https://download.openmmlab.com/mmcv/dist/cu102/torch1.6.0/index.html

    Please see here for different versions of MMCV compatible to different PyTorch and CUDA versions. Optionally you can choose to compile mmcv from source by the following command

    wget https://github.com/open-mmlab/mmcv/archive/refs/tags/v1.0.5.zip
unzip mmcv-1.0.5
cd mmcv
MMCV_WITH_OPS=1 pip install -e .  # package mmcv-full will be installed after this step
cd ..

    Or directly run

    pip install mmcv-full==1.0.5

  4. Please clone the MI-AOD repository.

    git clone https://github.com/yuantn/MI-AOD.git
cd MI-AOD

    If it is too slow, you can also try downloading the repository like this:

    wget https://github.com/yuantn/MI-AOD/archive/master.zip
unzip master.zip
cd MI-AOD-master

  5. Please install build requirements and then install and compile MMDetection in MI-AOD.

    pip install -r requirements/build.txt
pip install -v -e .  # or "python setup.py develop"

Please note:

a. Following the above instructions, MMDetection is installed on dev mode, any local modifications made to the code will take effect without the need to reinstall it.

b. If you would like to use opencv-python-headless instead of opencv-python, you can install it before installing MMCV.

c. Some dependencies are optional. Simply running pip install -v -e . will only install the minimum runtime requirements. To use optional dependencies like albumentations and imagecorruptions, either install them manually with pip install -r requirements/optional.txt, or specify desired extras when calling pip (e.g. pip install -v -e .[optional]). Valid keys for the extras field are: all, tests, build, and optional.

Modification in the MMCV Package

To train with two dataloaders (i.e., the labeled set dataloader and the unlabeled set dataloader mentioned in the paper) at the same time, you will need to modify the epoch_based_runner.py in the mmcv package.

Considering that this will affect all code that uses this environment, so we suggest you set up a separate environment for MI-AOD (i.e., the miaod environment created above).

cp -v epoch_based_runner.py $YOUR_ANACONDA_PATH/envs/miaod/lib/python3.7/site-packages/mmcv/runner/

Please change the $YOUR_ANACONDA_PATH to your actual Anaconda3 installation directory. Usually it would be ~/anaconda3.

After that, if you have modified anything in the mmcv package (including but not limited to: updating/re-installing Python, PyTorch, mmdetection, mmcv, mmcv-full, conda environment), you are supposed to copy the epoch_base_runner.py provided in this repository to the mmcv directory again. (Issue #3)

A From-scratch Setup Script

Assuming that you have Anaconda3 and CUDA 10.2 installed already, here is a full script for setting up MI-AOD with conda.

conda create -n miaod python=3.7 -y
conda activate miaod

conda install pytorch==1.6.0 torchvision==0.7.0 cudatoolkit=10.2 -c pytorch

# install the mmcv==1.0.5
pip install mmcv-full==1.0.5+torch1.6.0+cu102 -f https://download.openmmlab.com/mmcv/dist/index.html

# install mmdetection
git clone https://github.com/yuantn/MI-AOD.git
cd MI-AOD
pip install -r requirements/build.txt
pip install -v -e .  # or "python setup.py develop"

# modify epoch_based_runner.py
cp -v epoch_based_runner.py $YOUR_ANACONDA_PATH/envs/miaod/lib/python3.7/site-packages/mmcv/runner/

If you have any question, please feel free to leave an issue here, or refer to install.md in MMDetection V2.3.0.

And please refer to FAQ for frequently asked questions.