[Doc] Refine the doc of installation (#2546)

* [Doc] Refine the doc of installation
* add PULL_REQUEST_TEMPLATE.md

.github/PULL_REQUEST_TEMPLATE.md

@@ -0,0 +1,8 @@
+### PR types
+<!-- One of [ New features | Bug fixes | Function optimization | Performance optimization | Breaking changes | Others ] -->
+
+### PR changes
+<!-- One of [ Models | APIs | Docs | Others ] -->
+
+### Description
+<!-- Describe what this PR does -->

docs/install.md

@@ -1,7 +1,7 @@
 English | [简体中文](install_cn.md)
 
 
-## Environment Requirements
+## 1 Environment Requirements
 
 - PaddlePaddle (the version >= 2.2)
 - OS: 64-bit
@@ -10,9 +10,9 @@ English | [简体中文](install_cn.md)
 - CUDA >= 10.1
 - cuDNN >= 7.6
 
-## Installation
+## 2 Installation
 
-### 1. Install PaddlePaddle
+### 2.1 Install PaddlePaddle
 
 Please refer to the [installation doc](https://www.paddlepaddle.org.cn/documentation/docs/en/install/index_en.html) to install PaddlePaddle (the version >= 2.2).
 
@@ -39,48 +39,45 @@ In Python interpreter, run the following command to confirm whether PaddlePaddle
 
 ```
 
-### 2. Download Repo
+### 2.2 Install PaddleSeg
+
+If you make modification to `PaddleSeg/paddleseg`, e.g, adding model and loss, you should install PaddleSeg from source.
+
+If you only use PaddleSeg library, please install complied PaddleSeg.
+
+#### 2.2.1 Install PaddleSeg from Source
 
 Clone the PaddleSeg repo from Github.
 
 ```
 git clone https://github.com/PaddlePaddle/PaddleSeg
 ```
 
-### 3. Install Dependencies
+Run the following command, install PaddleSeg from source. If you make modification to `PaddleSeg/paddleseg`, it will be efficient without reinstallation.
 
 ```
 cd PaddleSeg
-pip install -r requirements.txt
+pip install -v -e .
 ```
+#### 2.2.2 Install Compiled PaddleSeg
 
-If a version error occurs during installation, you can try to delete the old version and re-run the script.
-### 4. Verify Installation
-
-In the root of PaddleSeg, run the following command. If there are no error in terminal log, you can use PaddleSeg to train, validate, test and export models with config method.
+Run the following command, install the complied PaddleSeg.
 
 ```
-cd PaddleSeg
-sh tests/run_check_install.sh
+pip install paddleseg
 ```
 
-### 5. Install PaddleSeg (Optional)
+### 2.3 Verify Installation
 
-Besides, PaddleSeg also supports api method for flexible development, which requires the PaddleSeg is installed in python libs.
-
-Run the following command to compile and install PaddleSeg from source code.
+In the root of PaddleSeg, run the following command.
+If there are no error in terminal log, you can use PaddleSeg to train, validate, test and export models with config method.
 
 ```
 cd PaddleSeg
-python setup.py install
-```
-
-Or, you can install the release PaddleSeg.
-```
-pip install paddleseg
+sh tests/install/check_predict.sh
 ```
 
-## Use PaddleSeg with Docker
+## 3 Use PaddleSeg with Docker
 
 Docker is an open-source tool to build, ship, and run distributed applications in an isolated environment. If you  do not have a Docker environment, please refer to [Docker](https://www.docker.com/). If you will use GPU version, you also need to install [nvidia-docker](https://github.com/NVIDIA/nvidia-docker).
 

docs/install_cn.md

@@ -2,7 +2,7 @@
 # 安装文档
 
 
-## 环境要求
+## 1 环境要求
 
 - PaddlePaddle (版本不低于2.2)
 - OS 64位操作系统
@@ -11,9 +11,9 @@
 - CUDA >= 10.1
 - cuDNN >= 7.6
 
-## 一、本地安装说明
+## 2 本地安装说明
 
-### 1. 安装PaddlePaddle
+### 2.1 安装PaddlePaddle
 
 请参考[快速安装文档](https://www.paddlepaddle.org.cn/install/quick)或者[详细安装文档](https://www.paddlepaddle.org.cn/documentation/docs/zh/install/index_cn.html)，安装PaddlePaddle （要求不低于2.2版本，推荐安装最新版本）。
 
@@ -38,7 +38,11 @@ python -m pip install paddlepaddle-gpu==2.2.2.post101 -f https://www.paddlepaddl
 
 ```
 
-### 2. 下载PaddleSeg代码
+### 2.2 安装PaddleSeg
+
+如果大家需要基于PaddleSeg进行开发和调试，推荐采用源码安装的方式。如果大家只是调用PaddleSeg，推荐安装发布的PaddleSeg包。
+
+#### 2.2.1 源码安装PaddleSeg
 
 从Github下载PaddleSeg代码。
 
@@ -52,45 +56,30 @@ git clone https://github.com/PaddlePaddle/PaddleSeg
 git clone https://gitee.com/paddlepaddle/PaddleSeg.git
 ```
 
-### 3. 安装PaddleSeg依赖
+执行如下命令，从源码编译安装PaddleSeg包。大家对于`PaddleSeg/paddleseg`目录下的修改，都会立即生效，无需重新安装。
 
 ```
 cd PaddleSeg
-pip install -r requirements.txt
+pip install -v -e .
 ```
 
-如果安装时出现版本错误，可以尝试删除旧版本，重新运行该脚本。
+#### 2.2.2 安装发布的PaddleSeg
 
-### 4.确认环境安装成功
-
-在PaddleSeg目录下执行如下命令，会进行简单的单卡训练和单卡预测。
+执行如下命令，安装发布的PaddleSeg包。
 
 ```
-sh tests/run_check_install.sh
+pip install paddleseg
 ```
 
-查看执行输出的log，没有报错，则验证安装成功。
-
-然后，基于PaddleSeg，你可以使用配置文件的方式来训练、验证、测试和导出模型。
-
-
-### 5. 安装PaddleSeg包（可选）
+### 2.3 确认环境安装成功
 
-PaddleSeg也支持API方式灵活开发。 API开发方式要求大家将PaddleSeg安装到Python库中。
+在PaddleSeg目录下执行如下命令，会进行简单的单卡预测。查看执行输出的log，没有报错，则验证安装成功。
 
-
-执行如下命令，从源码编译安装PaddleSeg包。
-```
-cd PaddleSeg
-python setup.py install
 ```
-
-或者直接安装发布的PaddleSeg包。
-```
-pip install paddleseg
+sh tests/install/check_predict.sh
 ```
 
-## 二、使用Docker快速体验PaddleSeg
+## 3 使用Docker快速体验PaddleSeg
 
 Docker是一种开源工具，用于在和系统本身环境相隔离的环境中构建、发布和运行各类应用程序。如果您没有Docker运行环境，请参考[Docker 官网](https://www.docker.com/)进行安装，如果您准备使用GPU版本镜像，还需要提前安装好[nvidia-docker](https://github.com/NVIDIA/nvidia-docker)。
 

tests/install/check_predict.sh

@@ -0,0 +1,6 @@
+# check predicting
+python predict.py \
+    --config configs/quick_start/pp_liteseg_optic_disc_512x512_1k.yml \
+    --model_path https://paddleseg.bj.bcebos.com/dygraph/optic_disc/pp_liteseg_optic_disc_512x512_1k/model.pdparams \
+    --image_path docs/images/optic_test_image.jpg \
+    --save_dir output/result