前言
本文记录一下在linux系安装PaddleDetection的过程，使用Conda的方式来安装；
（尝试过docker的方式，无法获取镜像；尝试过pip的方式，提升找不到库；最终使用Conda成功安装了。）
目录
一、设置国内源加速Conda
在Linux系统中，通过修改condarc文件，进行配置conda的源
vim ~/.condarc改为如下所示：
show_channel_urls: true
channel_alias: https://mirrors.tuna.tsinghua.edu.cn/anaconda
default_channels:
- https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
- https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free
- https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/r
- https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/pro
- https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/msys2
custom_channels:
conda-forge: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud
msys2: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud
bioconda: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud
menpo: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud
pytorch: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud
simpleitk: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud二、创建Conda环境
创建一个Conda环境，名称为PaddleDetection，指定python的版本为3.8
conda create -n PaddleDetection python=3.8
进入环境
conda activate PaddleDetection三、安装PaddlePaddle
首先来到官网，选择合适的CUDA版本，我选择了最新的CUDA11.7
根据上图的安装信息（命令），进行安装；
conda install paddlepaddle-gpu==2.4.2 cudatoolkit=11.7 -c https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/Paddle/ -c conda-forge然后配置环境变量（不配置的，在使用时会出现如下错误）
dynamic_loader.cc:307] The third-party dynamic library (libcudnn.so) that Paddle depends on is not configured correctly. (error code is /usr/local/cuda/lib64/libcudnn.so: cannot open shared object file: No such file or directory)
用conda env list，查询环境的路径
(PaddleDetection) root@bap3ac2457:/guopu # conda env list
# conda environments:
#
base /opt/conda
PaddleDetection * /opt/conda/envs/PaddleDetection
能看到刚创建的Conda的路径为 /opt/conda/envs/PaddleDetection
添加/opt/conda/envs/PaddleDetection/lib/到环境变量中.bashrc
echo "export LD_LIBRARY_PATH=/opt/conda/envs/PaddleDetection/lib/">>~/.bashrc测试是否安装成功
# 确认PaddlePaddle安装成功
python -c "import paddle; paddle.utils.run_check()"
# 确认PaddlePaddle版本
python -c "import paddle; print(paddle.__version__)"看到成功打印信息
(PaddleDetection) root@bap3ac2457:/guopu# python -c "import paddle; paddle.utils.run_check()"
Running verify PaddlePaddle program ...
W0508 08:48:03.937019 44515 gpu_resources.cc:61] Please NOTE: device: 0, GPU Compute Capability: 7.5, Driver API Version: 11.8, Runtime API Version: 11.7
W0508 08:48:03.950474 44515 gpu_resources.cc:91] device: 0, cuDNN Version: 8.4.
PaddlePaddle works well on 1 GPU.
PaddlePaddle works well on 1 GPUs.
PaddlePaddle is installed successfully! Let's start deep learning with PaddlePaddle now.
(PaddleDetection) root@bap3ac2457:/guopu#
四、安装PaddleDetection
首先下载PaddleDetection代码
git clone https://github.com/PaddlePaddle/PaddleDetection.git进入工程目录
cd PaddleDetection使用pip安装依赖，可以使用 -i https://pypi.tuna.tsinghua.edu.cn/simple 加速安装
pip install -r requirements.txt如果安装pycocotools报错了，参考https://blog.csdn.net/weixin_57096837/article/details/122775990
编译安装paddledet
python setup.py install如果出现下载一些库错误，比如pyclipper，那么手动安装：pip install pyperclip，再执行python setup.py install
如果出现error: protobuf 3.20.0 is installed but protobuf>=3.20.2 is required by {'onnx'}，执行：pip install protobuf==3.20.3
测试环境
安装后确认测试通过：
python ppdet/modeling/tests/test_architectures.py
测试通过后会提示如下信息：
(PaddleDetection) root@bap3ac2457:/guopu/PaddleDetection# python ppdet/modeling/tests/test_architectures.py
Warning: Unable to use numba in PP-Tracking, please install numba, for example(python3.7): `pip install numba==0.56.4`
Warning: Unable to use numba in PP-Tracking, please install numba, for example(python3.7): `pip install numba==0.56.4`
W0510 00:52:22.579213 47617 gpu_resources.cc:61] Please NOTE: device: 0, GPU Compute Capability: 7.5, Driver API Version: 11.8, Runtime API Version: 11.7
W0510 00:52:22.590267 47617 gpu_resources.cc:91] device: 0, cuDNN Version: 8.4.
.......
----------------------------------------------------------------------
Ran 7 tests in 3.731s
OK
(PaddleDetection) root@bap3ac2457:/guopu/PaddleDetection# 如果打印以上信息，说明安装完成啦~