Skip to content

rnasterofmysea/Taking_Medicine_Management

BranchesTags

Repository files navigation

Taking_Medicine_Management

Hardware Configuration

  • Raspbeery Pi 4
  • SD card 32G
  • element14 camera module

Server Configuration

  • Ubuntu 21.0.4 desktop verion
  • debian

Concept

Step 1 (item estimation)

  • 약통 인식 (실시간 스트리밍)

    • 약통 물체 인식
      • Yolov3 모델 > tenserflow 변환 후 구현
    • 약통에 적힌 숫자 인식
      • Tensorflow 사용

  • 흐름 Dialog

  • 약통 등록 상황
  1. 사용자가 카메라에 숫자가 적힌 약통을 보여줌
  2. tensorflow를 이용하여 약통과 약통에 적힌 숫자 인식
  3. 사용자에게 약통(약)에 대한 정보를 입력하도록 요구
  4. 약통(약) 정보와 인식된 숫자를 Spring을 통해 DB에 저장
  5. 약통 등록 완료
  • 약통 인식 상황
  1. 사용자가 카메라에 숫자가 적힌 (사전 등록된) 약통을 보여줌
  2. tensorflow를 이용하여 약통과 약통에 적힌 숫자 인식
  3. DB에 저장된 약통(약)에 대한 정보를 불러옴
  4. Step2로 넘어감

Step 2 (pose estimation)

ⓐ 환경구축

  1. 임베디드 환경 구축
  2. 우분투 서버 구축

ⓑ 데이터 추출 및 전처리

  1. Posenet 이미지 추출 페이지 ( 라즈베리파이 카메라)
  2. Posenet 이미지 --> 엑셀 추출 (3. 엑셀 라벨링)
  • Spring 기반 웹페이지

ⓒ AI 모델링

  1. Dense 모델 생성
  2. weight 학습
  3. 가중치 학습 결과 확인
  • Colab 사용

ⓓ 동영상 스트리밍 환경 구축

  1. Spring 웹페이지
  2. 라즈베리파이 스트리밍 API
  3. 동작인식 AI
  4. 포드포워딩 및 웹 서비스 구축

Proceed

Hardware & Software Setting

Raspberrypi4 SSH && Camera enable

sudo raspi-config
Interfacing Option > camera & ssh > enable

image

Static IP Assignment

  • check IP
ifconfig

image

  • IP assignment
 sudo vim /etc/dhcpcd.conf

image

VNC Connection

  • Raspberry PI setting
sudo raspi-config
Interfacing Option > VNC > enable
  • Install Ultra VNC

https://www.realvnc.com/en/connect/download/viewer/

image

  • Input IP Info & Login

image

image

image

Python Version Downgrading

!! To use tensorflow & OpenCV in raspberrypi4, you should downgrade python version and manage dependency --- python3.7.12

  • Check current version & path
which python
python -V

which python3
python3 -V

image

  • Firmware upgrade & update
sudo apt-get update & sudo apt-get upgrade
  • Install python 3.7.12
wget https://www.python.org/ftp/python/3.7.12/Python-3.7.12.tgz

sudo tar xzf Ptyhon-3.7.12.tgz

image

cd Python-3.7.21
./configure --enable-optimizations
sudo make -j 4
sudo make altinstall
  • Change default python version
echo "alias python=/usr/local/bin/python3.7" >> ~/.bashrc
source ~/.bashrc
  • Symbolic link python3.7.12 path
sudo ln -sf /usr/local/bin/python3.7 /usr/bin/python
sudo ln -sf /usr/local/bin/python3.7 /usr/bin/python3

image

  • Symblic link pip3.7 (to avoid dependency issue)

image

sudo ln -sf /usr/local/bin/pip3.7 /usr/bin/pip
sudo ln -sf /usr/local/bin/pip3.7 /usr/bin/pip3

Install Tensorflow

!! you should check firmware & python version and install suitable tensorflow

tensorflow-2.1.0-cp37-cp37m-linux_armv7l.whl

  • Check System
uname -a

image

  • Find & Check matching version

image

  • Make virtual envrionment
cd Desktop
mkdir project
cd project

python3 -m pip install virtualenv
python3 -m virtualenv env
source env/bin/activate

(type 'deactivate' to quit)

  • Install prerequired package
sudo apt-get install -y libhdf5-dev libc-ares-dev libeigen3-dev gcc gfortran libgfortran5 libatlas3-base libatlas-base-dev libopenblas-dev libopenblas-base libblas-dev liblapack-dev cython3 libatlas-base-dev openmpi-bin libopenmpi-dev python3-dev

pip install -U wheel mock six
sudo -H pip3 install --upgrade setuptools
  • Install tensorflow (tensorflow-2.1.0-cp37-cp37m-linux_armv7l.whl)
wget https://github.com/Qengineering/Tensorflow-Raspberry-Pi/raw/master/tensorflow-2.1.0-cp37-cp37m-linux_armv7l.whl

sudo -H pip3 install tensorflow-2.1.0-cp37-cp37m-linux_armv7l.whl

sudo reboot
  • Test tensorflow
python 
import tensorflow
tensorflow.__version__

image

Install OpenCV4

https://github.com/rnasterofmysea/RaspberryPi4_Yolov3

  • raspberrypi upgrade
sudo apt-get -y update && sudo apt-get -y  upgrade
sudo apt-get -y install python3-dev
  • install python package
pip3 install opencv-python
  • install opencv library
pip3 install opencv-contrib-python 
sudo apt-get install -y libatlas-base-dev libhdf5-dev libhdf5-serial-dev libatlas-base-dev libjasper-dev  libqtgui4  libqt4-test

or

sudo apt-get install libatlas-base-dev

Test OpenCV

  • make test.py
import cv2
img = cv2.imread("/lenna.png")

cv2.imshow("Test",img)

img_canny = cv2.Canny(img, 50, 150)

cv2.imshow("Test img Edge", img_canny)

cv2.waitKey(0)

cv2.destroyAllWindows()
  • save image in current directory

image

  • run test.py
python test.py

Next Cloud

Package upgrade & update & install

sudo apt upgrade
sudo apt update
sudo apt install nginx mariadb-server php php-fpm php-mysql php-zip php-common php-zip php-xml php-mbstring php-gd php-curl -y

Check packages

  • check raspberry pi IP address
ifconfig
  • connect [IP address]

image

Database setting

  • Enter mariadb as root
sudo mariadb -u root
  • Create database
MariaDB [(none)]> CREATE DATABASE nextcloud;
  • Create new account
MariaDB [(none)]> CREATE USER 'nextcloud'@'localhost' IDENTIFIED BY '1234';
  • Grant setting
MariaDB [(none)]> GRANT ALL PRIVILEGES ON nextcloud.* TO 'nextcloud'@'localhost';

quit

image

Install NextCloud

  • Download zip file
wget https://download.nextcloud.com/server/releases/latest.zip
  • unzip zip file
sudo rm /var/www/html/*
sudo unzip ./latest.zip -d /var/www/html/
sudo chown -R www-data:www-data /var/www/html
(auth setting)

Nginx setting

  • php 버전 확인
php -v

image

sudo nano /etc/nginx/sites-enabled/default
  • delete code > copy & paste
upstream php-handler {
    server unix:/var/run/php/php8.1-fpm.sock;
}

server {
    listen 80;
    listen [::]:80;
    server_name localhost;

    add_header X-Content-Type-Options nosniff;
    add_header X-XSS-Protection "1; mode=block";
    add_header X-Robots-Tag none;
    add_header X-Download-Options noopen;
    add_header X-Permitted-Cross-Domain-Policies none;
    add_header Referrer-Policy no-referrer;
    fastcgi_hide_header X-Powered-By;

    root /var/www/html/nextcloud;

    location = /robots.txt {
        allow all;
        log_not_found off;
        access_log off;
    }

    location = /.well-known/carddav {
      return 301 $scheme://$host:$server_port/remote.php/dav;
    }
    location = /.well-known/caldav {
      return 301 $scheme://$host:$server_port/remote.php/dav;
    }

    client_max_body_size 512M;
    fastcgi_buffers 64 4K;

    gzip on;
    gzip_vary on;
    gzip_comp_level 4;
    gzip_min_length 256;
    gzip_proxied expired no-cache no-store private no_last_modified no_etag auth;
    gzip_types application/atom+xml application/javascript application/json application/ld+json application/manifest+json application/rss+xml applicaEnter this intion/vnd.geo+json application/vnd.ms-fontobject application/x-font-ttf application/x-web-app-manifest+json application/xhtml+xml application/xml font/opentype image/bmp image/svg+xml image/x-icon text/cache-manifest text/css text/plain text/vcard text/vnd.rim.location.xloc text/vtt text/x-component text/x-cross-domain-policy;

    location / {
        rewrite ^ /index.php;
    }

    location ~ ^\/(?:build|tests|config|lib|3rdparty|templates|data)\/ {
        deny all;
    }
    location ~ ^\/(?:\.|autotest|occ|issue|indie|db_|console) {
        deny all;
    }

    location ~ ^\/(?:index|remote|public|cron|core\/ajax\/update|status|ocs\/v[12]|updater\/.+|oc[ms]-provider\/.+)\.php(?:$|\/) {
        fastcgi_split_path_info ^(.+?\.php)(\/.*|)$;
        set $path_info $fastcgi_path_info;
        try_files $fastcgi_script_name =404;
        include fastcgi_params;
        fastcgi_read_timeout 1800;
        fastcgi_param SCRIPT_FILENAME $document_root$fastcgi_script_name;
        fastcgi_param PATH_INFO $path_info;
        fastcgi_param modHeadersAvailable true;
        fastcgi_param front_controller_active true;
        fastcgi_pass php-handler;
        fastcgi_intercept_errors on;
        fastcgi_request_buffering off;
    }

    location ~ ^\/(?:updater|oc[ms]-provider)(?:$|\/) {
        try_files $uri/ =404;
        index index.php;
    }

    location ~ \.(?:css|js|woff2?|svg|gif|map)$ {
        try_files $uri /index.php$request_uri;
        add_header Cache-Control "public, max-age=15778463";
        add_header X-Content-Type-Options nosniff;
        add_header X-XSS-Protection "1; mode=block";
        add_header X-Robots-Tag none;
        add_header X-Download-Options noopen;
        add_header X-Permitted-Cross-Domain-Policies none;
        add_header Referrer-Policy no-referrer;

        access_log off;
    }

    location ~ \.(?:png|html|ttf|ico|jpg|jpeg|bcmap)$ {
        try_files $uri /index.php$request_uri;
        access_log off;
    }
}
  • Reload nginx
sudo nginx -s reload

Check Nginx

  • nginx 상태 확인
service nginx status

image

http://127.0.0.1/

http://localhost/

image

nextcloud와 로컬PC 연동

  • 로컬PC의 nextlcoud 실행 후 프로필 => 설정

image

  • 동기화 폴더 연결 추가

image

  • 동기화 할 로컬 폴더 선택

image

  • nextcloud의 원격 대상 폴더 선택

image

  • 동기화하지 않을 원격 폴더 선택

image

  • nextcloud와 로컬 PC 연동 완료

image

  • 동기화 확인

image

이미지 라벨링

  • ./windows_v1.8.1/data
    • 클래스를 추가할 수 있는 txt파일
  • ./windows_v1.8.1/labelImg.exe
    • 실행파일

image

predefined_classes.txt

  • medicine box를 추가

image

실행화면

  • 1번 클릭하여 YOLO로 변경
  • 2번 클릭 후 라벨링할 이미지 파일을 선택

image

이미지가 추가 된 화면

image

이미지 라벨링 시작

  1. 키보드 w클릭하여 이미지 라벨링
  2. txt파일에서 추가시킨 class명을 선택
  3. 저장하여 txt파일 추가
  4. 키보드 d를 눌러 다음 이미지로 전환
  5. 학습시킬 이미지 파일에 이동하여 classes.txt 파일 삭제

image

image

  • txt파일에는 이미지의 라벨링 좌표값이 저장되어있다.

image

이미지 라벨링 참고

Colab에서 라벨링한 이미지로 YOLO3 학습

1번

image

2번

image

3번

image

4번

image

5번

image

6번

image

학습된 YOLO3 나온 결과_1

image

학습된 YOLO3 나온 결과_2

image

YOLO3 파일 정보

  • yolov3_testing.cfg : YOLO3 config 파일
  • yolov3_training_last.weights : 학습된 YOLO3 가중치 데이터 image
import cv2
import numpy as np
from google.colab.patches import cv2_imshow

net = cv2.dnn.readNet("../gdrive/MyDrive/yolov3/yolov3_training_last.weights", "cfg/yolov3_training.cfg")
classes = []
with open("./data/coco.names", "r") as f:
    classes = [line.strip() for line in f.readlines()]
layer_names = net.getLayerNames()
output_layers = [layer_names[i - 1] for i in net.getUnconnectedOutLayers()] 
colors = np.random.uniform(0, 255, size=(len(classes), 3))
img = cv2.imread("test_medicine2.jpg")
img = cv2.resize(img, None, fx=0.4, fy=0.4)
height, width, channels = img.shape
blob = cv2.dnn.blobFromImage(img, 0.00392, (416, 416), (0, 0, 0), True, crop=False)
net.setInput(blob)
outs = net.forward(output_layers)
class_ids = []
confidences = []
boxes = []
flag = "flag"
for out in outs:
    for detection in out:
        scores = detection[5:]
        class_id = np.argmax(scores)
        confidence = scores[class_id]
        if confidence > 0.5:
            flag = str(confidence)
            center_x = int(detection[0] * width)
            center_y = int(detection[1] * height)
            w = int(detection[2] * width)
            h = int(detection[3] * height)            
            x = int(center_x - w / 2)
            y = int(center_y - h / 2)
            boxes.append([x, y, w, h])
            confidences.append(float(confidence))
            class_ids.append(class_id)
            indexes = cv2.dnn.NMSBoxes(boxes, confidences, 0.5, 0.4)
            from google.colab.patches import cv2_imshow
font = cv2.FONT_HERSHEY_PLAIN
for i in range(len(boxes)):
    if i in indexes:
        x, y, w, h = boxes[i]
        label = str(classes[class_ids[i]])
        color = colors[i]
        cv2.rectangle(img, (x, y), (x + w, y + h), color, 2)
        cv2.putText(img, label+flag, (x, y + 30), font, 3, color, 3)
cv2_imshow(img)
cv2.waitKey(0)
cv2.destroyAllWindows()

Python Selenium Scrapter ( Ubuntu & Jupyter)

!Selenum latest Version (version4)

  • setting url path by "By"

Ubuntu Code

from selenium import webdriver
from selenium.webdriver.chrome.service import Service
from webdriver_manager.chrome import ChromeDriverManager
from selenium.webdriver.common.keys import Keys
from selenium.webdriver.common.by import By
import time
import urllib.request
import os

#def set_chrome_driver():
#    chrome_options = webdriver.ChromeOptions()
#    driver = webdriver.Chrome(service=Service(ChromeDriverManager().install()), options=chrome_options)
#    return driver

def createDirectory(directory):
    try:
        if not os.path.exists(directory):
            os.makedirs(directory)
    except OSError:
        print("Error: Failed to create the directory.")

def crawling_img(name):
#     driver = webdriver.Chrome("D:\chromedriver\chromedriver")
    options = webdriver.ChromeOptions()
    options.add_argument('--headless')
    options.add_argument('--no-sandbox')
    options.add_argument('--disable-dev-shm-usage')
    driver = webdriver.Chrome(executable_path="/usr/bin/chromedriver",options=options)
    #driver = set_chrome_driver()
    driver.get("https://www.google.co.kr/imghp?hl=ko&tab=wi&authuser=0&ogbl")
    elem = driver.find_element(By.NAME,"q")
    elem.send_keys(name)
    elem.send_keys(Keys.RETURN)

    #
    SCROLL_PAUSE_TIME = 1
    # Get scroll height
    last_height = driver.execute_script("return document.body.scrollHeight")  # 브라우저의 높이를 자바스크립트로 찾음
    while True:
        # Scroll down to bottom
        driver.execute_script("window.scrollTo(0, document.body.scrollHeight);")  # 브라우저 끝까지 스크롤을 내림
        # Wait to load page
        time.sleep(SCROLL_PAUSE_TIME)
        # Calculate new scroll height and compare with last scroll height
        new_height = driver.execute_script("return document.body.scrollHeight")
        if new_height == last_height:
            try:
                driver.find_element(BY.CSS_SELECTOR,".mye4qd").click()
            except:
                break
        last_height = new_height

    imgs = driver.find_elements(By.CSS_SELECTOR, ".rg_i.Q4LuWd")
    dir = "./before" + "/" + name + "1"
    createDirectory(dir) #폴더 생성
    count = 1
    for img in imgs:
        try:
            img.click()
            time.sleep(2)
            imgUrl = driver.find_element(By.XPATH,
                '/html/body/div[2]/c-wiz/div[3]/div[1]/div/div/div/div[1]/div[1]/span/div[1]/div[1]/div[1]/a[1]/div[1]/img').get_attribute(
                "src")
            path = "./before" + "/" + name + "1/"
            urllib.request.urlretrieve(imgUrl, path + name + str(count) + ".jpg")
            count = count + 1
            if count >= 1000:
                break
        except:
            pass
    driver.close()
idols = ["white pill bottle"]
for idol in idols:
    crawling_img(idol)

Jupyter Code

from selenium import webdriver
from selenium.webdriver.chrome.service import Service
from webdriver_manager.chrome import ChromeDriverManager
from selenium.webdriver.common.keys import Keys
from selenium.webdriver.common.by import By
import time
import urllib.request
import os

def set_chrome_driver():
    chrome_options = webdriver.ChromeOptions()
    driver = webdriver.Chrome(service=Service(ChromeDriverManager().install()), options=chrome_options)
    return driver

def createDirectory(directory):
    try:
        if not os.path.exists(directory):
            os.makedirs(directory)
    except OSError:
        print("Error: Failed to create the directory.")

def crawling_img(name):
#     driver = webdriver.Chrome("D:\chromedriver\chromedriver")
    driver = set_chrome_driver()
    driver.get("https://www.google.co.kr/imghp?hl=ko&tab=wi&authuser=0&ogbl")
    elem = driver.find_element(By.NAME,"q")
    elem.send_keys(name)
    elem.send_keys(Keys.RETURN)

    #
    SCROLL_PAUSE_TIME = 1
    # Get scroll height
    last_height = driver.execute_script("return document.body.scrollHeight")  # 브라우저의 높이를 자바스크립트로 찾음
    while True:
        # Scroll down to bottom
        driver.execute_script("window.scrollTo(0, document.body.scrollHeight);")  # 브라우저 끝까지 스크롤을 내림
        # Wait to load page
        time.sleep(SCROLL_PAUSE_TIME)
        # Calculate new scroll height and compare with last scroll height
        new_height = driver.execute_script("return document.body.scrollHeight")
        if new_height == last_height:
            try:
                driver.find_element(BY.CSS_SELECTOR,".mye4qd").click()
            except:
                break
        last_height = new_height

    imgs = driver.find_elements(By.CSS_SELECTOR, ".rg_i.Q4LuWd")
    dir = ".\idols" + "\\" + name
    createDirectory(dir) #폴더 생성
    count = 1
    for img in imgs:
        try:
            img.click()
            time.sleep(2)
            #저장할 이미지 경로 및 
            imgUrl = driver.find_element(By.XPATH,
                '/html/body/div[2]/c-wiz/div[3]/div[1]/div/div/div/div[1]/div[1]/span/div[1]/div[1]/div[1]/a[1]/div[1]/img').get_attribute(
                "src")
            path = ".\idols" + "\\" + name + "\\"
            urllib.request.urlretrieve(imgUrl, path + name + str(count) + ".jpg")
            count = count + 1
            if count >= 10:
                break
        except:
            pass
    driver.close()
idols = ["white pill bottle"]
for idol in idols:
    crawling_img(idol)

Yolov3 + Live Streaming service

import cv2
import numpy as np
from google.colab.patches import cv2_imshow
# 웹캠 신호 받기
VideoSignal = cv2.VideoCapture("http://182.226.36.189:6974/?action=stream")
# YOLO 가중치 파일과 CFG 파일 로드
YOLO_net = cv2.dnn.readNet("../gdrive/MyDrive/yolov3/yolov3_training_last.weights","cfg/yolov3_training.cfg")

# YOLO NETWORK 재구성
classes = []
with open("./data/coco.names", "r") as f:
    classes = [line.strip() for line in f.readlines()]
layer_names = YOLO_net.getLayerNames()
output_layers = [layer_names[i - 1] for i in net.getUnconnectedOutLayers()] 

while True:
    # 웹캠 프레임
    ret, frame = VideoSignal.read()
    h, w, c = frame.shape

    # YOLO 입력
    blob = cv2.dnn.blobFromImage(frame, 0.00392, (416, 416), (0, 0, 0),
    True, crop=False)
    YOLO_net.setInput(blob)
    outs = YOLO_net.forward(output_layers)

    class_ids = []
    confidences = []
    boxes = []

    for out in outs:

        for detection in out:

            scores = detection[5:]
            class_id = np.argmax(scores)
            confidence = scores[class_id]

            if confidence > 0.5:
                # Object detected
                center_x = int(detection[0] * w)
                center_y = int(detection[1] * h)
                dw = int(detection[2] * w)
                dh = int(detection[3] * h)
                # Rectangle coordinate
                x = int(center_x - dw / 2)
                y = int(center_y - dh / 2)
                boxes.append([x, y, dw, dh])
                confidences.append(float(confidence))
                class_ids.append(class_id)
                indexes = cv2.dnn.NMSBoxes(boxes, confidences, 0.45, 0.4)
                

    for i in range(len(boxes)):
        if i in indexes:
            x, y, w, h = boxes[i]
            label = str(classes[class_ids[i]])
            score = confidences[i]

            # 경계상자와 클래스 정보 이미지에 입력
            cv2.rectangle(frame, (x, y), (x + w, y + h), (0, 0, 255), 5)
            cv2.putText(frame, label, (x, y - 20), cv2.FONT_ITALIC, 0.5, 
            (255, 255, 255), 1)
    from google.colab.patches import cv2_imshow
    cv2_imshow(frame)

    if cv2.waitKey(100) > 0:
        break

Result

  • 1초당 10 프레임 (10fps)

  • 인식 확인

  • 보안 사항

    • 인식률, rectablge box 좌표 반환

<<<<<<< HEAD

[Python] Image Data Arguement + Copy text file

======= image

d9c1456dd340004d668d86f368162b7232ee19a0

import random
import numpy as np
import os
import cv2
import glob
from PIL import Image
import PIL.ImageOps
import copy
from random import randint
import shutil

#다음 변수를 수정하여 새로 만들 이미지 갯수를 정합니다.
num_augmented_images = 2
file_path = 'C:\\Users\\scwon\\Desktop\\StudyDir\\jupyter_study\\img\\'
#file_names = os.listdir(file_path)
#total_origin_image_num = len(file_names)
file_list = os.listdir(file_path)
file_names = [file for file in file_list if file.endswith(".jpg")]
total_origin_image_num = len(file_names)
augment_cnt = 3


def SaltPepper(img):
    # Getting the dimensions of the image
    if img.ndim > 2:  # color
        height, width, _ = img.shape
    else:  # gray scaleㅋ
        height, width = img.shape
 
    result = copy.deepcopy(img)
 
    # Randomly pick some pixels in the image
    # Pick a random number between height*width/80 and height*width/10
    number_of_pixels = randint(int(height * width / 100), int(height * width / 10))
 
    for i in range(number_of_pixels):
        # Pick a random y coordinate
        y_coord = randint(0, height - 1)
 
        # Pick a random x coordinate
        x_coord = randint(0, width - 1)
 
        if result.ndim > 2:
            result[y_coord][x_coord] = [randint(0, 255), randint(0, 255), randint(0, 255)]
        else:
            # Color that pixel to white
            result[y_coord][x_coord] = 255
 
    # Randomly pick some pixels in image
    # Pick a random number between height*width/80 and height*width/10
    for i in range(number_of_pixels):
        # Pick a random y coordinate
        y_coord = randint(0, height - 1)
 
        # Pick a random x coordinate
        x_coord = randint(0, width - 1)
 
        if result.ndim > 2:
            result[y_coord][x_coord] = [randint(0, 255), randint(0, 255), randint(0, 255)]
        else:
            # Color that pixel to white
            result[y_coord][x_coord] = 0
 
    return result


for i in range(0, total_origin_image_num):
    change_picture_index = i
    print(file_names[i])
    file_name = file_names[change_picture_index]
    
    origin_image_path = 'C:\\Users\\scwon\\Desktop\\StudyDir\\jupyter_study\\img\\' + file_name
    print(origin_image_path)
    image = cv2.imread(origin_image_path)
    
    #1. 이미지 노이즈 추가 전환
    print("blur")
    converted_image = cv2.medianBlur(image, 3) 
    cv2.imwrite(file_path + str(augment_cnt) + '.jpg',converted_image)
    # txt파일 증폭
    src = 'C:\\Users\\scwon\\Desktop\\StudyDir\\jupyter_study\\img\\' + str(i) + ".txt"
    dst = 'C:\\Users\\scwon\\Desktop\\StudyDir\\jupyter_study\\img\\' + str(augment_cnt) + ".txt"
    shutil.copy(src, dst)
    augment_cnt += 1
    
    print("gray")
    converted_image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
    cv2.imwrite(file_path + str(augment_cnt) + '.jpg',converted_image)
    # txt파일 증폭
    src = 'C:\\Users\\scwon\\Desktop\\StudyDir\\jupyter_study\\img\\' + str(i) + ".txt"
    dst = 'C:\\Users\\scwon\\Desktop\\StudyDir\\jupyter_study\\img\\' + str(augment_cnt) + ".txt"
    shutil.copy(src, dst)
    augment_cnt += 1
    
    #2. 이미지 RGB 전환
    print("cv2.COLOR_BGR2RGB")
    converted_image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
    cv2.imwrite(file_path + str(augment_cnt) + '.jpg',converted_image)
    # txt파일 증폭
    src = 'C:\\Users\\scwon\\Desktop\\StudyDir\\jupyter_study\\img\\' + str(i) + ".txt"
    dst = 'C:\\Users\\scwon\\Desktop\\StudyDir\\jupyter_study\\img\\' + str(augment_cnt) + ".txt"
    shutil.copy(src, dst)
    augment_cnt += 1
    
    #3. 이미지 LUV 전환
    print("cv2.COLOR_BGR2Luv")
    converted_image = cv2.cvtColor(image, cv2.COLOR_BGR2Luv)
    cv2.imwrite(file_path + str(augment_cnt) + '.jpg',converted_image)
    # txt파일 증폭
    src = 'C:\\Users\\scwon\\Desktop\\StudyDir\\jupyter_study\\img\\' + str(i) + ".txt"
    dst = 'C:\\Users\\scwon\\Desktop\\StudyDir\\jupyter_study\\img\\' + str(augment_cnt) + ".txt"
    shutil.copy(src, dst)
    augment_cnt += 1
    
    #4. 이미지 noise 추가
    print("noise")
    converted_image = SaltPepper(image)
    cv2.imwrite(file_path + str(augment_cnt) + '.jpg',converted_image)
    # txt파일 증폭
    src = 'C:\\Users\\scwon\\Desktop\\StudyDir\\jupyter_study\\img\\' + str(i) + ".txt"
    dst = 'C:\\Users\\scwon\\Desktop\\StudyDir\\jupyter_study\\img\\' + str(augment_cnt) + ".txt"
    shutil.copy(src, dst)
    augment_cnt += 1
    
    #5. 이미지 선명하게 표현
    kernel = np.array([[0, -1, 0],
                   [-1, 5, -1],
                   [0, -1, 0]])
    # 커널 적용 
    converted_image = cv2.filter2D(image, -1, kernel)
    cv2.imwrite(file_path + str(augment_cnt) + '.jpg',converted_image)
    # txt파일 증폭
    src = 'C:\\Users\\scwon\\Desktop\\StudyDir\\jupyter_study\\img\\' + str(i) + ".txt"
    dst = 'C:\\Users\\scwon\\Desktop\\StudyDir\\jupyter_study\\img\\' + str(augment_cnt) + ".txt"
    shutil.copy(src, dst)
    augment_cnt += 1
    
    #6. 이미지 HLS 전환
    print("HLS")
    converted_image = cv2.cvtColor(image,cv2.COLOR_BGR2HLS)
    cv2.imwrite(file_path + str(augment_cnt) + '.jpg',converted_image)
    # txt파일 증폭
    src = 'C:\\Users\\scwon\\Desktop\\StudyDir\\jupyter_study\\img\\' + str(i) + ".txt"
    dst = 'C:\\Users\\scwon\\Desktop\\StudyDir\\jupyter_study\\img\\' + str(augment_cnt) + ".txt"
    shutil.copy(src, dst)
    augment_cnt += 1

Colab에서 YOLO3 학습 참고 자료

Reference

Python Scrapter

Selenum

Ubuntu

Jupyter

Yolov3 + OpenCV4 + Tesseract OCR

Yolov3 + Flask + Background running

Posenet + Pose estimation

Posenet + Dataset << * Chanwon

Downgrade Python

Install tensorflow

Trouble Issues

(Raspberry Pi 4) SSH connection timeout - fixed

https://sote.tistory.com/249 https://m.blog.naver.com/PostView.naver?isHttpsRedirect=true&blogId=duehd88&logNo=20201603052

About

No description, website, or topics provided.

Resources

Readme
Activity

Stars

0 stars

Watchers

1 watching

Forks

0 forks
Report repository

Releases

No releases published

Packages

 
 
 

Contributors

Languages