Yolov5_RK3588

* ### Ubuntu

Install Ubuntu on your RK3588 device. (tested on Ubuntu 20.04 and OrangePi5/Firefly ROC RK3588S devices)

For installing Ubuntu on Firefly you can use their manual[[1]](https://wiki.t-firefly.com/en/ROC-RK3588S-PC/index.html)[[2]](https://en.t-firefly.com/doc/download/page/id/142.html).

For installing Ubuntu on OrangePi you can use their manual.

Or use ours README's for them (select the one below).

|OrangePi|Firefly| | :---: | :---: |

* ### FFMPEG

Install ffmpeg package for WebUI:

sudo apt-get update
    sudo apt-get install -y ffmpeg

And dependencies for WebUI:

sudo apt-get update
    # General dependencies
    sudo apt-get install -y python-dev pkg-config
# Library components     sudo apt-get install libavformat-dev libavcodec-dev libavdevice-dev \       libavutil-dev libswscale-dev libswresample-dev libavfilter-dev

Open .bashrc in nano text editor:

nano ~/.bashrc

At the end of file add next line:

export LDPRELOAD=$LDPRELOAD:/usr/lib/aarch64-linux-gnu/libffi.so.7

Save and close nano with sortcuts ctrl-o, Enter, ctrl-x

### Docker (Optional)*

For installing docker on RK3588 device you can use official docker docs or check our README_DOCKER.md

* ### From Docker hub

At first you need download docker image:

docker pull deathk9t/yolov5_rk3588:latest

Then you can run container with:

docker run --privileged --name [container-name] -e DISPLAY=$DISPLAY -v /tmp/.X11-unix:/tmp/.X11-unix -v /dev/:/dev --network host -it deathk9t/yolov5_rk3588:latest

* ### Build docker image by yourself

You can build docker image by yourself usning Dockerfile:

docker build -t [name-docker-image:tag] .

Then you can run container with:

docker run --privileged --name [container-name] -e DISPLAY=$DISPLAY -v /tmp/.X11-unix:/tmp/.X11-unix -v /dev/:/dev --network host -it [name-docker-image:tag]

Install miniconda

wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-aarch64.sh
  bash Miniconda3-latest-Linux-aarch64.sh

Then rerun terminal session:

source ~/.bashrc

Create conda env with python3.9

conda create -n <env-name> python=3.9

And then activate conda env

conda activate <env-name>

Clone repository:

git clone https://github.com/Applied-Deep-Learning-Lab/Yolov5_RK3588

And got into repo-dir:

cd Yolov5_RK3588

Install RKNN-Toolkit2-Lite，such as rknntoolkitlite2-1.4.0-cp39-cp39-linux_aarch64.whl

pip install install/rknntoolkitlite2-1.4.0-cp39-cp39-linux_aarch64.whl

In created conda enviroment also install requirements from the same directory

pip install -r install/requirements.txt

Then go to the install dir for building and installing cython_bbox

cd install/cython_bbox
  python3 setup.py build
  python3 setup.py install

`main.py` runs inference with WebUI. You can turn on/off some options in config file or using Settings page at webUI.

python3 main.py

Or run it using bash script:

source run.sh

The frame rate dropped by about 20 fps due to recording. When running without it, you can expect around 60 frames per second.

For see WebUI write to browser address bar next (localhost - device's ip):

localhost:8080

You also can set autostart for running this.

Before it deactivate conda env:

conda deactivate

* For Orange Pi

source install/autostart/orangepi_autostart.sh

* For Firefly:

source install/autostart/firefly_autostart.sh

* ### Host PC

Install Python3 and pip3

sudo apt-get update
      sudo apt-get install python3 python3-dev python3-pip

Install dependent libraries

sudo apt-get update
      sudo apt-get install libxslt1-dev zlib1g zlib1g-dev libglib2.0-0 libsm6 libgl1-mesa-glx libprotobuf-dev gcc git

Install RKNN-Toolkit2，such as rknntoolkit2-1.4.022dcfef4-cp38-cp38-linuxx8664.whl

pip install resources/HostPC/converter/install/rknntoolkit2-1.4.022dcfef4-cp38-cp38-linuxx8664.whl

For convert your .onnx model to .rknn run onnx2rknn.py like:

cd resources/HostPC/converter/convert/
      python3 onnx2rknn.py \
              --input <path-to-your-onnx-model> \
              --output <path-where-save-rknn-model> \
              --dataset <path-to-txt-file-with-calibration-images-names>