Skip to content

aimuch/multitask

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

7 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

multitasks based on openmmlab

文件树说明

.
├── configs # 配置文件
│   ├── det
│   ├── det3d
│   └── seg
├── demo    # 演示
│   ├── det
│   ├── det3d
│   └── seg
├── mtl # 多任务
│   ├── apis        # 接口
│   ├── datasets    # 数据集
│   ├── engine      # 引擎
│   ├── evaluation  # 评估
│   ├── __init__.py # 初始化
│   ├── model       # 模型
│   ├── registry    # 注册
│   ├── utils       # 工具
│   └── version.py  # 版本
├── plugin   # 插件
│   ├── __init__.p      # 初始化
│   ├── mmdet3d_plugin  # 3d检测源码
│   ├── mmdet_plugin    # 2d检测源码
│   └── mmseg_plugin    # 分割源码
├── projects    # 项目
│   ├── det     # 2d检测
│   ├── det3d   # 3d检测
│   └── seg     # 分割
├── README.md
├── requirements    # 依赖
├── requirements.txt
├── setup.py    # 安装
├── tests   # 测试
│   ├── det     # 2d检测
│   ├── det3d   # 3d检测
│   └── seg     # 分割
└── tool    # 工具
    ├── det     # 2d检测
    ├── det3d   # 3d检测
    └── seg     # 分割

Requirements

  • CUDA=11.8
  • Pytorch=2.0.0
  • mmcv=2.0.0
  • mmdet=2.0.0
  • mmseg=1.0.0

环境搭建

pytorch安装

conda install pytorch torchvision torchaudio pytorch-cuda=11.8 -c pytorch -c nvidia

mmcv和mmengine安装

可以通过mim安装或者pip安装,两种方式均可

  • mim安装
    pip install -U openmim
    mim install mmengine
    mim install mmcv==2.0.0
  • pip安装
    pip install mmcv==2.0.0 -f https://download.openmmlab.com/mmcv/dist/cu118/torch2.0/index.html

mmdet/mmdet3d/mmseg可以用plugin中的源码,也可以用pip安装或者官方源码安装

[TODO]

mmdet安装

可以通过mim安装或者源码安装,两种方式均可

  • pip安装
    如果你将 mmdet 作为依赖或第三方 Python 包
    mim install mmdet==2.0.0
  • 源码安装
    在需要修改源码时要用源码安装
    git clone https://github.com/open-mmlab/mmdetection.git
    cd mmdetection
    pip install -v -e .
    # "-v" 指详细说明,或更多的输出
    # "-e" 表示在可编辑模式下安装项目,因此对代码所做的任何本地修改都会生效,从而无需重新安装。
    这里可以用本仓库中的plugin/mmdet_plugin中的源码安装

mmdet3d安装

可以通过pip安装或者源码安装,两种方式均可

  • pip安装
    如果你将 mmdet 作为依赖或第三方 Python 包
    mim install "mmdet3d>=1.1.0"
  • 源码安装
    在需要修改源码时要用源码安装
    git clone https://github.com/open-mmlab/mmdetection3d.git
    cd mmdetection
    pip install -v -e .
    # "-v" 指详细说明,或更多的输出
    # "-e" 表示在可编辑模式下安装项目,因此对代码所做的任何本地修改都会生效,从而无需重新安装。
    这里可以用本仓库中的plugin/mmdet3d_plugin中的源码安装

mmseg安装

可以通过mim安装或者源码安装,两种方式均可

  • pip安装
    如果你将 mmseg 作为依赖或第三方 Python 包
    pip install "mmsegmentation>=1.0.0"
  • 源码安装
    在需要修改源码时要用源码安装
    git clone -b main https://github.com/open-mmlab/mmsegmentation.git
    cd mmsegmentation
    pip install -v -e .
    # '-v' 表示详细模式,更多的输出
    # '-e' 表示以可编辑模式安装工程,因此对代码所做的任何修改都生效,无需重新安装
    这里可以用本仓库中的plugin/mmseg_plugin中的源码安装

验证安装

python -c 'import torch;print(torch.__version__);print(torch.version.cuda);print(torch.cuda.is_available())'
python -c 'import mmcv;print(mmcv.__version__)'
python -c 'import mmdet;print(mmdet.__version__)'
python -c 'import mmdet;print(mmdet3d.__version__)'
python -c 'import mmseg;print(mmseg.__version__)'

自定义模型

项目结构

About

MTL, Multi-Task Learning

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages