OpenCV va Tensorflow yordamida Dragonboard 410c yoki 820c bilan ob'ektlarni aniqlash: 4 qadam

2024 Muallif: John Day | [email protected]. Oxirgi o'zgartirilgan: 2024-01-30 13:25

Bu ko'rsatmalarni Ob'ektlarni aniqlash dasturini ishga tushirish uchun Python 3.5 uchun OpenCV, Tensorflow va mashinani o'rganish ramkalarini qanday o'rnatishni tasvirlaydi.

1 -qadam: talablar

Sizga quyidagi belgilar kerak bo'ladi:

• DragonBoard ™ 410c yoki 820c;

• Linaro-alipning toza o'rnatilishi:

  • DB410c: v431 versiyasida sinovdan o'tgan. Havola:
  • DB820c: v228 versiyasida sinovdan o'tgan. Havola:
• Kamida 16 Gb hajmli MicroSD karta (agar 410c ishlatilsa);

Faylni yuklab oling (bu qadam oxirida), oching va MicroSD kartasiga nusxa ko'chiring; Obs: Agar DB820c dan foydalansangiz, faylni yuklang, oching va buyruqlardan foydalanishni osonlashtirish uchun/home/*USER*/ga o'ting.

• USB uyasi;
• USB kamera (Linux bilan mos keladi);
• USB sichqoncha va klaviatura;
• Internetga ulanish.

Obs: Agar iloji bo'lsa, bu ko'rsatmalarni DragonBoard brauzerida bajaring, bu esa buyruqlarni nusxalashni osonlashtiradi

2 -qadam: MicroSD kartasini o'rnatish (faqat W/ DB410c)

• Dragonboard -da terminalni oching;
• Terminalda fdiskni ishga tushiring:

$ sudo fdisk -l

• MicroSD kartasini DragonBoard MicroSD karta uyasiga joylashtiring;
• Ro'yxatdagi yangi qurilmaning nomini (va bo'limini) qidirib, fdiskni qayta ishga tushiring (masalan, mmcblk1p1)

$ sudo fdisk -l

Ildiz katalogiga o'ting:

$ CD ~

Jild yaratish:

$ mkdir sdfolder

MicroSD kartasini o'rnating:

$ mount / dev / sdfolder

3 -qadam: Kerakli ramkalarni o'rnatish

• Dragonboard -da terminalni oching;
• Terminalda tanlangan katalogga o'ting (820c uchun "~" va 410c uchun o'rnatilgan SDCard yordamida):

(820c) $ CD ~

(410c) $ cd ~/sdfolder

Ob'ektni aniqlash skriptlari papkasiga o'ting:

$ cd ob'ekt_detector_tensorflow_opencv/skriptlar/

Atrof -muhitni sozlash skriptini ishga tushiring:

$ sudo bash set_Env.sh

Tizimni yangilang:

$ sudo apt yangilash

Ushbu paketlarni o'rnating:

$ sudo apt install -y protobuf-compiler gcc-aarch64-linux-gnu

g ++-aarch64-linux-gnu debootstrap schroot git curl pkg-config zip unzip python python-pip g ++ zlib1g-dev default-jre libhdf5-dev libatlas-base-dev gfortran v4l-utils hdf5* libhdf5* libpng-dev build-essential libreadline-gplv2-dev libncursesw5-dev libssl-dev libsqlite3-dev tk-dev libgdbm-dev libc6-dev libbz2-dev libjpeg-dev libtiff5-dev libavcodec-dev libavformat-dev libsvscl-dev-libvxv2 libgtk2.0-dev libgtk-3-dev ffmpeg python-opengl

Ushbu katalogga o'ting:

$ cd /usr /src

Python 3.5 -ni yuklab oling:

$ sudo wget

Paketni chiqarib oling:

$ sudo tar xzf Python-3.5.6.tgz

Siqilgan paketni o'chirib tashlang:

$ sudo rm Python-3.5.6.tgz

Python 3.5 katalogiga o'ting:

$ CD Python-3.5.6

Python 3.5 kompilyatsiyasi uchun optimallashtirishni yoqing:

$ sudo./configure-faollashtirish-optimallashtirish

Python 3.5 ni kompilyatsiya qiling:

$ sudo altinstall qiling

Pip va sozlash vositalarini yangilang:

$ sudo python3.5 -m pip o'rnatish -pip && python3.5 -m pip o'rnatish -setuptools -ni yangilash

Numpy -ni o'rnating:

$ python3.5 -m pip o'rnatish numpy

Tanlangan katalogga o'ting:

(820c) $ CD ~

(410c) $ cd ~/sdfolder

Tensorflow 1.11 -ni yuklab oling:

$ wget

Tensorflowni o'rnating:

$ sudo python3.5 -m pip o'rnatish tensorflow-1.11.0-cp35-none-linux_aarch64.whl

OpenCV va OpenCV Contrib omborlarini klonlang:

$ sudo git clone -b 3.4 https://github.com/opencv/opencv.git && sudo git clone -b 3.4

Katalogga o'ting:

$ CD ochiq

Qurilish katalogini yarating va unga o'ting:

$ sudo mkdir build && cd build

CMake -ni ishga tushiring:

$ sudo cmake -D CMAKE_BUILD_TYPE = RELEASE -D CMAKE_INSTALL_PREFIX =/usr/local -D BUILD_opencv_java = OFF -D BUILD_opencv_python = OFF -D BUILD_opencv_python3 = qaysi python3.5) -D PYTHON_INCLUDE_DIR =/usr/local/include/python3.5m/-D INSTALL_C_EXAMPLES = O'chirilgan -D INSTALL_PYTHON3_EXAMPLES = O'chirilgan -D BUILD_EXAMPLES = O'CHIRILGAN -BU -OFF = BT -BU -OFF = -DBUILD_TBB = ON -D OPENCV_ENABLE_NONFREE = ON -DBUILD_opencv_xfeatures2d = OFF -D OPENGL = ON -D OPENMP = ON -D ENABLE_NEON = ON -D BUILD_PERF_TESTS = OFF/ON -ON/ON -ON modullar..

OpenCV -ni 4 yadroli kompilyatsiya qiling:

$ sudo make -j 4

OpenCV -ni o'rnating:

$ sudo make install

Tanlangan katalogga o'ting:

(820c) $ CD ~

(410c) $ cd ~/sdfolder

Skriptlar katalogiga o'ting:

$ cd ob'ekt_detector_tensorflow_opencv/skriptlar/

Python3.5 talablarini o'rnating:

$ sudo python3.5 -m pip install -r requirements.txt --no -cache -dir

Sinov importi:

$ python 3.5

> cv2 import >> import tensorflow

Obs: Agar cv2 import xatosini qaytarsa, OpenCV tuzish papkasida make install -ni ishga tushiring va qaytadan urinib ko'ring

Tanlangan katalogga o'ting:

(820c) $ CD ~

(410c) $ cd ~/sdfolder

Cocoapi omborini yuklab oling:

$ git klon

Tensorflow modellari omborini yuklab oling:

$ git klon

Ushbu katalogga o'ting:

$ cd cocoapi/PythonAPI

3 va 8 -qatorda pythonni python3.5 ga o'zgartirib, Makefile faylini tahrirlang va faylni saqlang (misol sifatida nanodan foydalaning):

$ nano Makefile

Kokoapini kompilyatsiya qiling:

$ sudo qilish

Obs: Agar "make" buyrug'i tuzilmasa, cython -ni qayta o'rnatishga harakat qiling:

$ sudo python3.5 -m pip cython o'rnatish

Pikokotoollarni tensorflow /modellar /tadqiqot katalogiga nusxalash:

(820c) $ cp -r pycocotools ~/modellar/tadqiqot/

(410c) $ cp -r pycocotools ~/sdfolder/modellar/tadqiqot/

Tanlangan katalogga o'ting:

(820c) $ CD ~

(410c) $ cd ~/sdfolder

Modellar/tadqiqot katalogiga o'ting:

$ CD modellari/tadqiqotlari

Protoc bilan kompilyatsiya qiling:

$ protoc object_detection/protos/*. proto --python_out =.

Eksport muhiti o'zgaruvchisi:

$ export PYTHONPATH = $ PYTHONPATH: `pwd`:` pwd`/slim

Atrof -muhitni sinab ko'ring:

$ python3.5 object_detection/quruvchilar/model_builder_test.py

Obs: OK qaytarilishi kerak, aks holda dastur ishlamaydi. Agar yo'q bo'lsa, kerakli ramkalarni o'rnatish jarayonida xatolarni diqqat bilan qidiring

4 -qadam: Ob'ektlarni aniqlash API -ni ishga tushirish

Barcha tuzilmalar sozlangan holda, endi Tensorflow bilan birga OpenCV -dan foydalanadigan ob'ektlarni aniqlash API -ni ishga tushirish mumkin.

Tanlangan katalogga o'ting:

(820c) $ CD ~

(410c) $ cd ~/sdfolder

Ob'ektlarni aniqlash katalogiga o'ting:

$ cd ob'ekt_detector_tensorflow_opencv/

Endi dasturni ishga tushiring:

$ python3.5 app.py

Endi Dragonboard videoni tarmoq orqali uzatadi. Chiqish videosini ko'rish uchun brauzerni JBda oching va "0.0.0.0: 5000" ga o'ting.