### Install cuDNN 8.1 Samples

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/test_tipc/docs/install.md

Installs cuDNN 8.1 runtime, developer, and samples libraries using dpkg. It then copies the samples to the home directory and compiles the mnistCUDNN example to verify the installation.

```bash
# x.x.x表示下载的版本号
# $HOME为工作目录
sudo dpkg -i libcudnn8_x.x.x-1+cudax.x_arm64.deb
sudo dpkg -i libcudnn8-dev_8.x.x.x-1+cudax.x_arm64.deb
sudo dpkg -i libcudnn8-samples_8.x.x.x-1+cudax.x_arm64.deb

# 验证是否正确安装
cp -r /usr/src/cudnn_samples_v8/ $HOME
cd  $HOME/cudnn_samples_v8/mnistCUDNN

# 编译
make clean && make
./mnistCUDNN
```

--------------------------------

### Install Python Requirements

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/deploy/third_engine/demo_avh/README.md

Install the necessary Python libraries for the demo. Run this command from the directory containing the requirements.txt file.

```bash
pip install -r ./requirements.txt
```

--------------------------------

### Install LabelMe on Ubuntu

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/docs/tutorials/data/DetAnnoTools_en.md

Installs LabelMe using apt-get, pip, or a standalone executable on Ubuntu systems.

```bash
sudo apt-get install labelme

# or
sudo pip3 install labelme

# or install standalone executable from:
# https://github.com/wkentaro/labelme/releases
```

--------------------------------

### Install PaddleDet

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/deploy/auto_compression/README.md

Install the PaddleDet library for object detection models.

```shell
pip install paddledet
```

--------------------------------

### Install Paddle Serving Packages

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/deploy/serving/cpp/README.md

Install the necessary Paddle Serving client, server, and application packages. Choose the appropriate server package (CPU or GPU) for your environment. Consider using a domestic mirror for faster downloads.

```commandline
pip install paddle-serving-client
# pip install paddle-serving-server # CPU
pip install paddle-serving-server-gpu # GPU 默认 CUDA10.2 + TensorRT6，其他环境需手动指定版本号
pip install paddle-serving-app
# pip install paddlepaddle # CPU
pip install paddlepaddle-gpu
```

--------------------------------

### Start Training and Evaluation

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/docs/tutorials/data/PrepareDetDataSet.md

Execute this command to start training your model with the specified configuration and enable evaluation during training. Ensure CUDA is set if using GPU.

```bash
export CUDA_VISIBLE_DEVICES=0
python tools/train.py -c configs/yolov3/yolov3_mobilenet_v1_roadsign.yml --eval
```

--------------------------------

### Install VisualDL

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/docs/tutorials/QUICK_STARTED.md

Installs VisualDL for real-time visualization of training metrics. Requires Python 3.5 or later.

```python
python -m pip install visualdl -i https://mirror.baidu.com/pypi/simple
```

--------------------------------

### Basic CMake Setup

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/deploy/third_engine/demo_ncnn/CMakeLists.txt

Sets the minimum CMake version and C++ standard. Initializes the project name.

```cmake
cmake_minimum_required(VERSION 3.9)
set(CMAKE_CXX_STANDARD 17)

project(picodet_demo)
```

--------------------------------

### Install PaddlePaddle

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/deploy/auto_compression/README.md

Install PaddlePaddle for CPU or GPU environments.

```shell
# CPU
pip install paddlepaddle
# GPU
pip install paddlepaddle-gpu
```

--------------------------------

### Install PaddleDetection Dependencies and Package

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/docs/tutorials/INSTALL.md

Installs the necessary dependencies for PaddleDetection using requirements.txt and then compiles and installs the PaddleDetection package itself. This should be run from the cloned PaddleDetection directory.

```shell
cd PaddleDetection
pip install -r requirements.txt
python setup.py install
```

--------------------------------

### Install ONNX Runtime

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/deploy/EXPORT_ONNX_MODEL.md

Install the ONNX Runtime library for performing inference with ONNX models.

```bash
pip install onnxruntime
```

--------------------------------

### Install LabelImg on Ubuntu

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/docs/tutorials/data/DetAnnoTools_en.md

Install necessary packages and dependencies for LabelImg on Ubuntu using apt-get and pip.

```bash
sudo apt-get install pyqt5-dev-tools
sudo pip3 install -r requirements/requirements-linux-python3.txt
make qt5py3
python3 labelImg.py
python3 labelImg.py [IMAGE_PATH] [PRE-DEFINED CLASS FILE]
```

--------------------------------

### Install VisualDL

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/docs/tutorials/logging_en.md

Install the VisualDL library using pip. This is required for logging metrics to VisualDL.

```bash
pip install visualdl
```

--------------------------------

### YOLOv3 Configuration File Example

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/docs/advanced_tutorials/MODEL_TECHNICAL_en.md

This is an example of a general configuration file for YOLOv3, demonstrating how different modules like dataset, runtime, optimizer, network structure, and reader are included and can be overridden.

```yaml
_BASE_: [
  '../datasets/coco_detection.yml', # Dataset configuration file shared by all models
  '../runtime.yml', # Runtime configuration
  '_base_/optimizer_270e.yml', # Optimizer related configuration
  '_base_/yolov3_darknet53.yml', # yolov3 Network structure configuration file
  '_base_/yolov3_reader.yml', # yolov3 Reader module configuration
]

# The relevant configuration defined here can override the configuration of the same name in the above file
snapshot_epoch: 5
weights: output/yolov3_darknet53_270e_coco/model_final

```

--------------------------------

### Install ONNX Prerequisites for Model Conversion

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/deploy/third_engine/demo_openvino/README.md

Navigate to the model_optimizer directory and install the necessary requirements for ONNX model conversion.

```shell
cd <INSTSLL_DIR>/openvino_2021/deployment_tools/model_optimizer
cd ./install_prerequisites
sudo install_prerequisites_onnx.sh
```

--------------------------------

### Install Paddle2ONNX and ONNX

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/configs/rtdetr/README.md

Install necessary libraries for converting models to ONNX format.

```shell
pip install onnx==1.13.0
pip install paddle2onnx==1.0.5
```

--------------------------------

### Compile and Install Paddle from Source

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/test_tipc/docs/install.md

Clones the PaddlePaddle repository, configures the build with specific options (MKL, GPU, distributed, Release, TensorRT), compiles it, and installs the GPU version.

```bash
# 编译安装Paddle
git clone https://github.com/PaddlePaddle/Paddle.git
cd Paddle
mkdir build && cd build
cmake .. \
  -DWITH_MKL=ON \
  -DWITH_MKLDNN=ON \
  -DWITH_GPU=ON \
  -DWITH_DISTRIBUTE=ON \
  -DCMAKE_BUILD_TYPE=Release \
  -DCUDA_ARCH_NAME=Auto \
  -DPY_VERSION=3.7 \
  -DON_INFER=ON \
  -DWITH_TENSORRT=ON \
  -DTENSORRT_ROOT=/usr/local/TensorRT6-cuda10.1-cudnn7
make -j 20
pip3.7 install python/dist/paddlepaddle_gpu-0.0.0-cp37-cp37m-linux_x86_64.whl
cd ../..

```

--------------------------------

### Install LabelImg with Anaconda

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/docs/tutorials/data/DetAnnoTools_en.md

Install LabelImg using Anaconda, including PyQt and lxml, and generate resource files.

```bash
conda install pyqt=5
conda install -c anaconda lxml
pyrcc5 -o libs/resources.py resources.qrc
python labelImg.py
python labelImg.py [IMAGE_PATH] [PRE-DEFINED CLASS FILE]
```

--------------------------------

### Download and Unzip Data Example

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/docs/tutorials/data/PrepareDetDataSet_en.md

Command to navigate to the dataset directory and download/unzip data, typically used for preparing custom datasets.

```bash
# Downing and unziping data
  >>cd $(ppdet_root)/dataset

```

--------------------------------

### Install Weights & Biases SDK

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/docs/tutorials/logging_en.md

Install the wandb SDK and log in to your wandb account to use Weights & Biases for experiment tracking.

```bash
pip install wandb
wandb login
```

--------------------------------

### Start Training with VisualDL Visualization

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/docs/tutorials/GETTING_STARTED_cn.md

This command starts the training process for a YOLOv3 model and enables VisualDL for real-time monitoring of loss and mAP trends. Ensure CUDA_VISIBLE_DEVICES is set if using a GPU on Linux/Mac.

```bash
export CUDA_VISIBLE_DEVICES=0 #windows和Mac下不需要执行该命令
python tools/train.py -c configs/yolov3/yolov3_mobilenet_v1_roadsign.yml
                        --use_vdl=true \
                        --vdl_log_dir=vdl_dir/scalar \

```

--------------------------------

### Install LabelMe on macOS

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/docs/tutorials/data/DetAnnoTools_en.md

Installs LabelMe using Homebrew and pip, or a command-line interface/app via Homebrew on macOS.

```bash
brew install pyqt  # maybe pyqt5
pip install labelme

# or
brew install wkentaro/labelme/labelme  # command line interface
# brew install --cask wkentaro/labelme/labelme  # app

# or install standalone executable/app from:
# https://github.com/wkentaro/labelme/releases
```

--------------------------------

### Install FreeImage Library

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/test_tipc/docs/install.md

Installs the freeimage library, which may be required if the mnistCUDNN sample fails with freeimage-related errors.

```bash
sudo apt-get install libfreeimage-dev
sudo apt-get install libfreeimage
```

--------------------------------

### Install PaddleSlim from Source

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/configs/slim/README_en.md

Compile and install PaddleSlim from its source code. This method is useful for development or when specific build configurations are needed.

```shell
git clone https://github.com/PaddlePaddle/PaddleSlim.git
cd PaddleSlim
python setup.py install
```

--------------------------------

### Start LabelImg

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/docs/tutorials/data/DetAnnoTools_en.md

Launch LabelImg using a Python script, optionally specifying an image path.

```bash
python labelImg.py <IMAGE_PATH>
```

--------------------------------

### VOC Data Format: train.txt Example

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/docs/tutorials/data/PrepareDetDataSet_en.md

This example illustrates the format of the 'train.txt' file, which lists the paths to training images and their corresponding annotation files.

```bash
# train.txt is list of trainset
>>cat train.txt
./images/xxx1.jpg ./annotations/xxx1.xml
./images/xxx2.jpg ./annotations/xxx2.xml
...
```

--------------------------------

### Include Directories Setup

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/deploy/third_engine/demo_ncnn/CMakeLists.txt

Adds necessary include directories for OpenCV, the current source directory, and the current binary directory.

```cmake
include_directories(
    ${OpenCV_INCLUDE_DIRS}
    ${CMAKE_CURRENT_SOURCE_DIR}
    ${CMAKE_CURRENT_BINARY_DIR}
)
```

--------------------------------

### YOLOv3 Reader Configuration Example

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/docs/advanced_tutorials/READER.md

An example configuration for a YOLOv3 reader, defining worker number, sample transforms, batch transforms, batch size, shuffle, and drop_last settings for training, evaluation, and testing.

```yaml
worker_num: 2
TrainReader:
  sample_transforms:
    - Decode: {}
    ...
  batch_transforms:
    ...
  batch_size: 8
  shuffle: true
  drop_last: true
  use_shared_memory: true

EvalReader:
  sample_transforms:
    - Decode: {}
    ...
  batch_size: 1
  drop_empty: false

TestReader:
  inputs_def:
    image_shape: [3, 608, 608]
  sample_transforms:
    - Decode: {}
    ...
  batch_size: 1

```

--------------------------------

### Build PicoDet Demo (Windows)

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/deploy/third_engine/demo_openvino/README.md

Build the PicoDet demo application on Windows using CMake and MSBuild. Ensure OpenVINO environment variables are set.

```cmd
<OPENVINO_INSTSLL_DIR>\openvino_2021\bin\setupvars.bat
mkdir -p build
cd build
cmake ..
msbuild picodet_demo.vcxproj /p:configuration=release /p:platform=x64
```

--------------------------------

### CMake Build Configuration Example (No TensorRT)

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/deploy/cpp/docs/Jetson_build.md

An example configuration for CMake build on Jetson, specifically disabling TensorRT integration. This setup is useful when TensorRT is not required or not available.

```bash
# Whether to use GPU (i.e., use CUDA)
WITH_GPU=ON

# Whether to use MKL or openblas
WITH_MKL=OFF

# Whether to integrate TensorRT (only valid when WITH_GPU=ON)
WITH_TENSORRT=OFF

# TensorRT's include path
TENSORRT_INC_DIR=/usr/include/aarch64-linux-gnu

# TensorRT's lib path
TENSORRT_LIB_DIR=/usr/lib/aarch64-linux-gnu

# Paddle inference library path
PADDLE_DIR=/home/nvidia/PaddleDetection_infer/paddle_inference/

# Paddle inference library name
PADDLE_LIB_NAME=paddle_inference

# Whether to compile the Paddle inference library using static libraries
# When using TensorRT, Paddle's inference library is usually a dynamic library
WITH_STATIC_LIB=OFF

# CUDA lib path
CUDA_LIB=/usr/local/cuda-10.0/lib64

# CUDNN lib path
CUDNN_LIB=/usr/lib/aarch64-linux-gnu/

# Whether to enable keypoint model prediction function
WITH_KEYPOINT=ON
```

--------------------------------

### Prepare Data and Environment

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/test_tipc/docs/benchmark_train.md

Run this script to prepare training data and install necessary environments. It takes a configuration file and a mode as arguments.

```shell
bash test_tipc/prepare.sh test_tipc/configs/faster_rcnn/faster_rcnn_r50_fpn_1x_coco_train_infer_python.txt benchmark_train
```

--------------------------------

### CMake Build Configuration for Jetson

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/deploy/cpp/docs/Jetson_build.md

Configure CMake variables for building the Paddle prediction library on Jetson. Ensure paths for TensorRT, Paddle inference library, CUDA, CUDNN, and OpenCV are correctly set according to your system setup. This example shows a configuration without TensorRT.

```bash
# Whether to use GPU (i.e., use CUDA)
WITH_GPU=ON

# Whether to use MKL or openblas, TX2 needs to be set to OFF
WITH_MKL=OFF

# Whether to integrate TensorRT (only valid when WITH_GPU=ON)
WITH_TENSORRT=OFF

# TensorRT's include path
TENSORRT_INC_DIR=/usr/include/aarch64-linux-gnu

# TensorRT's lib path
TENSORRT_LIB_DIR=/usr/lib/aarch64-linux-gnu

# Paddle inference library path
PADDLE_DIR=/path/to/paddle_inference/ 

# Paddle inference library name
PADDLE_LIB_NAME=paddle_inference

# Whether to compile the Paddle inference library using static libraries
# When using TensorRT, Paddle's inference library is usually a dynamic library
WITH_STATIC_LIB=OFF

# CUDA lib path
CUDA_LIB=/usr/local/cuda-10.0/lib64

# CUDNN lib path
CUDNN_LIB=/usr/lib/aarch64-linux-gnu

# Whether to enable keypoint model prediction function
WITH_KEYPOINT=ON

# OPENCV_DIR path
# For Linux platforms, please download: https://bj.bcebos.com/paddleseg/deploy/opencv3.4.6gcc4.8ffmpeg.tar.gz2 and decompress it into the deps folder
# For TX2 platforms, please download: https://paddlemodels.bj.bcebos.com/TX2_JetPack4.3_opencv_3.4.10_gcc7.5.0.zip and decompress it into the deps folder
OPENCV_DIR=/path/to/opencv

# Please check if all the above paths are correct

# The following does not need to be changed
cmake .. \
    -DWITH_GPU=${WITH_GPU} \
    -DWITH_MKL=OFF \
    -DWITH_TENSORRT=${WITH_TENSORRT} \
    -DTENSORRT_DIR=${TENSORRT_DIR} \
    -DPADDLE_DIR=${PADDLE_DIR} \
    -DWITH_STATIC_LIB=${WITH_STATIC_LIB} \
    -DCUDA_LIB=${CUDA_LIB} \
    -DCUDNN_LIB=${CUDNN_LIB} \
    -DOPENCV_DIR=${OPENCV_DIR} \
    -DPADDLE_LIB_NAME={PADDLE_LIB_NAME} \
    -DWITH_KEYPOINT=${WITH_KEYPOINT}
```

--------------------------------

### Install LabelMe with Anaconda

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/docs/tutorials/data/DetAnnoTools_en.md

Installs LabelMe using Anaconda, creating a dedicated environment and installing necessary packages.

```bash
conda create –name=labelme python=3
conda activate labelme
pip install pyqt5
pip install labelme
```

--------------------------------

### Install AutoLog for PaddleDetection

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/test_tipc/docs/install.md

Clones the AutoLog repository, installs its requirements, builds the wheel, and then installs the wheel package. This is a dependency for PaddleDetection.

```bash
# 安装AutoLog
git clone https://github.com/LDOUBLEV/AutoLog
cd AutoLog
pip3.7 install -r requirements.txt
python3.7 setup.py bdist_wheel
pip3.7 install ./dist/auto_log-1.0.0-py3-none-any.whl

# 下载PaddleDetection代码
cd ../
git clone https://github.com/PaddlePaddle/PaddleDetection
```

--------------------------------

### Install PaddleDetection and TIPC Requirements

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/test_tipc/docs/test_train_inference_python.md

Installs necessary dependencies for PaddleDetection and its TIPC testing framework. Ensure PaddlePaddle >= 2.2 is installed prior to running these commands.

```shell
pip install -r ./requirements.txt
pip install -r ./test_tipc/requirements.txt
```

--------------------------------

### Run PicoDet Demo (Inference Video)

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/deploy/third_engine/demo_openvino/README.md

Run the PicoDet demo to perform inference on a video file. Replace VIDEO_PATH with the actual path to your video file.

```shell
picodet_demo 2 VIDEO_PATH
```

--------------------------------

### Run PicoDet Demo (Benchmark)

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/deploy/third_engine/demo_openvino/README.md

Run the PicoDet demo in benchmark mode to measure performance. The argument '0' typically indicates the device to use for benchmarking.

```shell
picodet_demo 3 0
```

--------------------------------

### Verify PaddlePaddle Installation

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/docs/tutorials/INSTALL.md

Commands to import PaddlePaddle and run a check to confirm a successful installation. Also includes a command to print the installed PaddlePaddle version.

```python
>>> import paddle
>>> paddle.utils.run_check()
```

```python
python -c "import paddle; print(paddle.__version__)"
```

--------------------------------

### Build PicoDet Demo (Linux)

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/deploy/third_engine/demo_openvino/README.md

Build the PicoDet demo application on Linux using CMake and make. Ensure OpenVINO environment variables are sourced.

```shell
source /opt/intel/openvino_2021/bin/setupvars.sh
mkdir build
cd build
cmake ..
make
```

--------------------------------

### Install pycocotools for Windows

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/docs/tutorials/INSTALL.md

Provides an alternative installation command for pycocotools, specifically for Windows users where the default installation might fail. This version supports Python 3.

```shell
pip install git+https://github.com/philferriere/cocoapi.git#subdirectory=PythonAPI
```

--------------------------------

### Run PicoDet Demo (Webcam)

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/deploy/third_engine/demo_openvino/README.md

Run the PicoDet demo to perform inference on a live webcam feed. The first argument '0' indicates the webcam device.

```shell
picodet_demo 0 0
```

--------------------------------

### Install Cython for Older Python Versions

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/docs/tutorials/INSTALL.md

Installs the cython package, which may be required for pycocotools installation on Python versions 3.6 or earlier, to resolve potential Distutils errors.

```shell
pip install cython
```

--------------------------------

### Start Client to Access Serving

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/deploy/serving/cpp/README.md

Initiate the client to send requests to the running prediction service. Provide the path to the serving client configuration, the image file for inference, and the HTTP port the server is listening on.

```commandline
python deploy/serving/python/serving_client.py --serving_client output_inference/yolov3_darknet53_270e_coco/serving_client --image_file demo/000000014439.jpg --http_port 9997
```

--------------------------------

### Install AutoLog for Standardized Logging

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/test_tipc/docs/test_train_inference_python.md

Clones the AutoLog repository, installs its dependencies, builds the wheel, and then installs the wheel package. This tool is used for normalizing log output during TIPC tests.

```shell
git clone https://github.com/LDOUBLEV/AutoLog
cd AutoLog
pip install -r ./requirements.txt
python setup.py bdist_wheel
pip install ./dist/auto_log-1.0.0-py3-none-any.whl
```

--------------------------------

### Start Paddle Serving Service (CPU)

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/deploy/serving/README.md

Launches the Paddle Serving server on CPU. Ensure the model files are in the current directory.

```python
cd output_inference/yolov3_darknet53_270e_coco/

# CPU
python -m paddle_serving_server.serve --model serving_server --port 9393
```

--------------------------------

### Install MNN Python Library

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/deploy/third_engine/demo_mnn_kpts/README.md

Install the MNN Python library using pip.

```shell
pip install MNN
```

--------------------------------

### YOLOv5 FPGM Quantization Training Configuration

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/test_tipc/configs/yolov5/yolov5_l_300e_coco_train_infer_python.txt

Start training with FPGM quantization. This command uses the base configuration and a specific slim configuration for FPGM.

```bash
tools/train.py -c configs/yolov5/yolov5_l_300e_coco.yml --slim_config _template_fpgm -o
```

--------------------------------

### Start Docker Image for Paddle

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/test_tipc/docs/install.md

Launches a Docker container with the PaddlePaddle image, mapping the current directory to /paddle. This is the recommended way to set up the environment.

```bash
# 启动docker镜像
nvidia-docker run --name paddle -it -v $PWD:/paddle paddlepaddle/paddle:latest-gpu-cuda10.1-cudnn7-gcc82-dev /bin/bash
cd /paddle
```

--------------------------------

### Install PaddleYOLO

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/README_en.md

Clone the repository and install dependencies. Requires Python 3.7.0+ and PaddlePaddle 2.3.2+.

```bash
git clone https://github.com/PaddlePaddle/PaddleYOLO  # clone
cd PaddleYOLO
pip install -r requirements.txt  # install
```

--------------------------------

### PicoDet Training Configuration

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/test_tipc/benchmark_train.txt

This snippet shows the command to start a normal training process for the ppyolo_mbv3_large_coco model using a configuration file.

```bash
tools/train.py -c configs/ppyolo/ppyolo_mbv3_large_coco.yml -o
```

--------------------------------

### Start Training with Custom Configuration

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/docs/advanced_tutorials/customization/detection_en.md

Execute the training script with your modified configuration file. The --eval flag enables evaluation during training.

```bash
export CUDA_VISIBLE_DEVICES=0
python tools/train.py -c configs/yolov3/yolov3_mobilenet_v1_270e_coco.yml --eval
```

--------------------------------

### Run Demo with Local FVP

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/deploy/third_engine/demo_avh/README.md

Run the demo application on local Fixed Virtual Platforms (FVPs) by specifying the --enable_FVP argument. This is an alternative to using the Arm Virtual Hardware AMI instance.

```bash
./run_demo.sh --enable_FVP 1
```

--------------------------------

### Install ADB on macOS

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/deploy/lite/README.md

Install the Android Debug Bridge (ADB) tool on macOS using Homebrew.

```shell
brew cask install android-platform-tools
```

--------------------------------

### Start Paddle Serving Service (GPU)

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/deploy/serving/README.md

Launches the Paddle Serving server on a GPU. Ensure the model files are in the current directory.

```python
cd output_inference/yolov3_darknet53_270e_coco/

# GPU
python -m paddle_serving_server.serve --model serving_server --port 9393 --gpu_ids 0
```

--------------------------------

### Install PaddleDetection Dependencies

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/test_tipc/docs/install.md

Navigates into the PaddleDetection directory and installs its required Python packages using the requirements.txt file.

```bash
cd PaddleDetection
pip3.7 install -r ./requirements.txt
```

--------------------------------

### Configure PicoDet Demo for Inference

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/deploy/third_engine/demo_openvino/README.md

Edit the main.cpp and picodet_openvino.h files to set the image size and the path to the OpenVINO model weights.

```c++
step1:
main.cpp
#define image_size 416
...
auto detector = PicoDet("../weight/picodet_m_416.xml");
...
step2:
picodet_openvino.h
#define image_size 416
```

--------------------------------

### Run PicoDet Demo (Inference Images)

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/deploy/third_engine/demo_openvino/README.md

Run the PicoDet demo to perform inference on a batch of images located in a specified folder. Replace IMAGE_FOLDER with the actual path to your images.

```shell
picodet_demo 1 IMAGE_FOLDER/*.jpg
```

--------------------------------

### Install LabelImg on macOS

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/docs/tutorials/data/DetAnnoTools_en.md

Install LabelImg on macOS using Homebrew or pip, including Qt and lxml dependencies.

```bash
brew install qt  # Install qt-5.x.x by Homebrew
brew install libxml2

or using pip

pip3 install pyqt5 lxml # Install qt and lxml by pip

make qt5py3
python3 labelImg.py
python3 labelImg.py [IMAGE_PATH] [PRE-DEFINED CLASS FILE]
```

--------------------------------

### Edit Main C++ File for Demo

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/deploy/third_engine/demo_openvino_kpts/README.md

Update the `main.cpp` file to set the image size, initialize the image matrix, define center and scale, and specify the paths to the downloaded OpenVINO models for PicoDet and TinyPose.

```cpp
#define image_size 416
...
  cv::Mat image(256, 192, CV_8UC3, cv::Scalar(1, 1, 1));
  std::vector<float> center = {128, 96};
  std::vector<float> scale = {256, 192};
...
  auto detector = PicoDet("../weight/picodet_m_416.xml");
  auto kpts_detector = new KeyPointDetector("../weight/tinypose256.xml", -1, 256, 192);
...
```

--------------------------------

### Install PaddleSlim via pip

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/configs/slim/README_en.md

Install PaddleSlim directly using pip. This is the simplest method for adding PaddleSlim to your project.

```shell
pip install paddleslim -i https://pypi.tuna.tsinghua.edu.cn/simple
```

--------------------------------

### Run the Demo Application

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/deploy/third_engine/demo_avh/README.md

Execute the main script to run the bare metal text recognition application. This command initiates the demo process.

```bash
./run_demo.sh
```

--------------------------------

### Train a Model with Compression

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/configs/slim/README_en.md

Start the training process for a model, applying specified compression configurations. Ensure you have both the model configuration and the slim configuration files ready.

```shell
python tools/train.py -c configs/{MODEL.yml} --slim_config configs/slim/{SLIM_CONFIG.yml}
```

--------------------------------

### PicoDet Quantized Training Configuration

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/test_tipc/benchmark_train.txt

This command initiates training with quantization-aware training (QAT) for the ppyolo_mbv3_large_coco model, utilizing a specific slim configuration.

```bash
tools/train.py -c configs/ppyolo/ppyolo_mbv3_large_coco.yml --slim_config configs/slim/quant/ppyolo_mbv3_large_qat.yml -o
```

--------------------------------

### Install PaddlePaddle CPU version

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/docs/tutorials/INSTALL.md

Installs the CPU-only version of PaddlePaddle, version 2.3.2. This is suitable for environments without a compatible GPU.

```shell
python -m pip install paddlepaddle==2.3.2 -i https://mirror.baidu.com/pypi/simple
```

--------------------------------

### Install ADB on Linux

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/deploy/lite/README.md

Install the Android Debug Bridge (ADB) tool on Linux systems using apt package manager.

```shell
sudo apt update
sudo apt install -y wget adb
```

--------------------------------

### Install Paddle Lite with pip

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/deploy/lite/README.md

Install the Paddle Lite library using pip. This is the recommended method for obtaining the paddle_lite_opt tool.

```shell
pip install paddlelite
```

--------------------------------

### Run TinyPose MNN Demo for Benchmark

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/deploy/third_engine/demo_mnn_kpts/README.md

Execute the compiled demo to run a speed benchmark. Mode '3' is for benchmark, and the second argument is typically '0' for a dummy input.

```shell
./tinypose-mnn "3" "0"
```

--------------------------------

### COCO Dataset Image Information Example

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/docs/tutorials/data/PrepareDetDataSet.md

This is an example of the information structure for a single image within a COCO annotation JSON file.

```json
{
    'license': 3,
    'file_name': '000000391895.jpg',
    'coco_url': 'http://images.cocodataset.org/train2017/000000391895.jpg',
    'height': 360,
    'width': 640,
    'date_captured': '2013-11-14 11:18:45',
    'flickr_url': 'http://farm9.staticflickr.com/8186/8119368305_4e622c8349_z.jpg',
    'id': 391895
}
```

--------------------------------

### Launch VisualDL Server

Source: https://github.com/paddlepaddle/paddleyolo/blob/develop/docs/tutorials/GETTING_STARTED_cn.md

This command starts the VisualDL server to view training logs. It defaults to running on 127.0.0.1 but can be configured with a different IP using the --host parameter.

```shell
# 下述命令会在127.0.0.1上启动一个服务，支持通过前端web页面查看，可以通过--host这个参数指定实际ip地址
visualdl --logdir vdl_dir/scalar/

```