cysin / MobileNet-SSD-1

Ultra-fast MobileNet-SSD + Neural Compute Stick(NCS) than YoloV2 + Explosion speed by RaspberryPi · Multiple moving object detection with high accuracy.

Home Page:https://qiita.com/PINTO/items/b97b3334ed452cb555e2

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

[Japanese] MobileNet-SSD

YoloV2 より超速 MobileNetSSD+Neural Compute Stick(NCS)+Raspberry Piによる爆速・高精度の複数動体検知
映像再生と物体検出は非同期実行。
マルチスティックを実現するために、マルチスレッド かつ OpenGL で実装している。
【Qiita記事】 https://qiita.com/PINTO/items/b97b3334ed452cb555e2
【RealSense D435対応版】 https://github.com/PINTO0309/MobileNet-SSD-RealSense.git

動作イメージ

MobileNet-SSD + Neural Compute Stick + RaspberryPi3 / MultiStick(3本/Hard)

Youtube: https://youtu.be/sQnFbRSqIA8

Riders MultiStick

環境

・RaspberryPi 3 + Raspbian Stretch

・NCSDK v1.12.00 (NCSDK 2.04 では動作しません)

・Intel Movidius Neural Compute Stick 1本

・OpenCV 3.4.1

・OpenGL

・numpy

・UVC対応のUSB-Webカメラ

環境構築

  1. SWAP領域の一時的な拡張
$ sudo nano /etc/dphys-swapfile
CONF_SWAPSIZE=2048

$ sudo /etc/init.d/dphys-swapfile restart swapon -s
  1. パッケージのインストール
$ sudo apt-get update
$ sudo apt-get upgrade
$ sudo apt-get install python3-pip python3-numpy git cmake
  1. NCSDKのインストール
$ cd ~
$ git clone https://github.com/movidius/ncsdk.git
$ cd ncsdk
$ make install
  1. OpenCVのインストール
$ wget https://github.com/PINTO0309/OpenCVonARMv7/raw/master/libopencv3_3.4.1-20180304.1_armhf.deb
$ sudo apt install -y ./libopencv3_3.4.1-20180304.1_armhf.deb
$ sudo ldconfig
  1. OpenGLのインストール
$ sudo apt-get install python-opengl
$ sudo -H pip3 install pyopengl
$ sudo -H pip3 install pyopengl_accelerate
$ sudo raspi-config
  1. 「7.Advanced Options」-「A7 GL Driver」-「G2 GL (Fake KMS)」の順に選択し、Raspberry Pi のOpenGL Driver を有効化

  2. 再起動

$ sudo reboot
  1. リソース一式のダウンロード
$ cd ~
$ git clone https://github.com/PINTO0309/MobileNet-SSD.git
  1. USB-WEBカメラ(UVC対応) と Neural Compute Stick をRaspberryPiのUSBポートへ接続(Neural Compute Stickをマルチで使用する場合は電圧が不足するためセルフパワーUSB-Hub必須)

  2. RaspberryPiとディスプレイをHDMIケーブルで接続

  3. MobileNet-SSDの実行

$ cd MobileNet-SSD
$ python3 MultiStickSSD.py
  1. SWAP領域の縮小
$ sudo nano /etc/dphys-swapfile
CONF_SWAPSIZE=100

$ sudo /etc/init.d/dphys-swapfile restart swapon -s

独自学習データの生成手順

https://github.com/movidius/ncappzoo/tree/master/caffe/SSD_MobileNet
https://github.com/FreeApe/VGG-or-MobileNet-SSD
https://github.com/chuanqi305/MobileNet-SSD

[English] MobileNet-SSD

Ultra-fast MobileNet-SSD + Neural Compute Stick(NCS) than YoloV2 + Explosion speed by RaspberryPi.
Multiple moving object detection with high accuracy.
Video playback and object detection are executed asynchronously.
In order to realize high speed rendering with multi stick, it is implemented in multithreading/OpenGL.

【Japanese Article】https://qiita.com/PINTO/items/b97b3334ed452cb555e2
【RealSense D435 Version】 https://github.com/PINTO0309/MobileNet-SSD-RealSense.git

Image of motion

MobileNet-SSD + Neural Compute Stick + RaspberryPi3 / MultiStick(3 Stick / Hard Motion)

Youtube: https://youtu.be/sQnFbRSqIA8

Riders MultiStick

Environment

・RaspberryPi 3 + Raspbian Stretch

・NCSDK v1.12.00 (It does not work with NCSDK v2.04)

・Intel Movidius Neural Compute Stick 1 piece

・OpenCV 3.4.1

・OpenGL

・numpy

・(UVC)USB-Web Camera

Building environment

  1. Temporary extension of SWAP area
$ sudo nano /etc/dphys-swapfile
CONF_SWAPSIZE=2048

$ sudo /etc/init.d/dphys-swapfile restart swapon -s
  1. Installing packages
$ sudo apt-get update
$ sudo apt-get upgrade
$ sudo apt-get install python3-pip python3-numpy git cmake
  1. Installing NCSDK
$ cd ~
$ git clone https://github.com/movidius/ncsdk.git
$ cd ncsdk
$ make install
  1. Installation of OpenCV
$ wget https://github.com/PINTO0309/OpenCVonARMv7/raw/master/libopencv3_3.4.1-20180304.1_armhf.deb
$ sudo apt install -y ./libopencv3_3.4.1-20180304.1_armhf.deb
$ sudo ldconfig
  1. Installing OpenGL
$ sudo apt-get install python-opengl
$ sudo -H pip3 install pyopengl
$ sudo -H pip3 install pyopengl_accelerate
$ sudo raspi-config
  1. 「7.Advanced Options」-「A7 GL Driver」-「G2 GL (Fake KMS)」 and Activate Raspberry Pi's OpenGL Driver

  2. Reboot

$ sudo reboot
  1. Download complete set of resources
$ cd ~
$ git clone https://github.com/PINTO0309/MobileNet-SSD.git
  1. Connect USB-WEB camera (UVC compatible) and Neural Compute Stick to RaspberryPi's USB port (self power USB-Hub required due to insufficient voltage when using Neural Compute Stick in multiple)

  2. Connect RaspberryPi and display with HDMI cable

  3. Running MobileNet-SSD

$ cd MobileNet-SSD
$ python3 MultiStickSSD.py
  1. Reducing the SWAP area
$ sudo nano /etc/dphys-swapfile
CONF_SWAPSIZE=100

$ sudo /etc/init.d/dphys-swapfile restart swapon -s

Procedure for generating original learning data

https://github.com/movidius/ncappzoo/tree/master/caffe/SSD_MobileNet
https://github.com/FreeApe/VGG-or-MobileNet-SSD
https://github.com/chuanqi305/MobileNet-SSD

About

Ultra-fast MobileNet-SSD + Neural Compute Stick(NCS) than YoloV2 + Explosion speed by RaspberryPi · Multiple moving object detection with high accuracy.

https://qiita.com/PINTO/items/b97b3334ed452cb555e2

License:MIT License


Languages

Language:Python 100.0%