PaddleOCR使用笔记-快速安装

PaddleOCR使用笔记

Linux环境下

快速安装

• 首先查看一下自己的cuda版本（10.1.168）

cat /usr/local/cuda/version.txt

• 再看一下cuDNN的版本（7.6.1）

cat /usr/local/cuda/include/cudnn.h | grep CUDNN_MAJOR -A 2

• 安装好docker和nvidia-docker后【参考教程：https://lexiaoyuan.blog.csdn.net/article/details/118053600、https://lexiaoyuan.blog.csdn.net/article/details/118054085】，在https://www.paddlepaddle.org.cn/install/quick?docurl=/documentation/docs/zh/2.0/install/docker/linux-docker.html中选择对应的选项，复制下面的命令。

• 拉取预安装PaddlePaddle的镜像：

nvidia-docker pull registry.baidubce.com/paddlepaddle/paddle:2.0.0-gpu-cuda10.1-cudnn7

• 用镜像构建并进入docker容器：

cd /home/lexiaoyuan/projects # 进入自己的存放项目的目录，因为下面的命令需要把当前目录挂载到docker容器中

nvidia-docker run --name paddle -it -v $PWD:/paddle registry.baidubce.com/paddlepaddle/paddle:2.0.0-gpu-cuda10.1-cudnn7 /bin/bash

• 使用git，下载PaddleOCR（下载最新的dygraph分支的）到本地，在把PaddleOCR上传到服务器。

git clone https://gitee.com/paddlepaddle/PaddleOCR.git --branch dygraph

• 由于默认的python3是指向python3.5，我们把它改成指向python3.8

ls -l /usr/bin | grep python

cd /usr/bin
rm python3
y
ln -s python3.8 python3

• 上面这个修改办法不管用的话，就用这个：

echo alias python3=python3.8 >> ~/.bashrc
source ~/.bashrc

• 进入docker中的paddle容器里的PaddleOCR目录,并安装第三方库

cd /paddle/PaddleOCR
pip3.8 install -r requirements.txt -i https://pypi.tuna.tsinghua.edu.cn/simple

• Ok,到这里，PaddleOCR的运行环境就装好了。

---

补充：安装nccl2

官方文档：https://docs.nvidia.com/deeplearning/nccl/install-guide/index.html

2.4.8： https://docs.nvidia.com/deeplearning/nccl/archives/nccl_248/nccl-install-guide/index.html

2.73: https://docs.nvidia.com/deeplearning/nccl/archives/nccl_273/install-guide/index.html

https://www.paddlepaddle.org.cn/documentation/docs/zh/install/compile/linux-compile.html

https://developer.nvidia.com/nccl/nccl-legacy-downloads

• 步骤如下：

In order to download NCCL, ensure you are registered for the NVIDIA Developer Program.

1. Go to: NVIDIA NCCL home page.
2. Click Download.
3. Complete the short survey and click Submit.
4. Accept the Terms and Conditions. A list of available download versions of NCCL displays.
5. Select the NCCL version you want to install. A list of available resources displays. Refer to the following sections to choose the correct package depending on the Linux distribution you are using.

• Install the keys.

sudo apt-key adv --fetch-keys http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64/7fa2af80.pub

sudo wget https://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu1604/x86_64/nvidia-machine-learning-repo-ubuntu1604_1.0.0-1_amd64.deb

sudo dpkg -i nvidia-machine-learning-repo-ubuntu1604_1.0.0-1_amd64.deb

sudo apt update

sudo apt install -y libnccl2=2.4.8-1+cuda10.1 libnccl-dev=2.4.8-1+cuda10.1

sudo apt remove libnccl2=2.4.8-1+cuda10.1 libnccl-dev=2.4.8-1+cuda10.1

sudo apt install -y libnccl2=2.7.3-1+cuda10.1 libnccl-dev=2.7.3-1+cuda10.1

• 测试nccl

文档：https://github.com/NVIDIA/nccl-tests/

https://zhuanlan.zhihu.com/p/174710896

git clone https://github.com/NVIDIA/nccl-tests.git
cd nccl-tests
make
./build/all_reduce_perf -b 8 -e 128M -f 2 -g 8

创作不易，喜欢的话加个关注点个赞，❤谢谢谢谢❤