從子文件夾中讀取所有圖像，檢測文本並將它們與原始名稱和文件夾一起保存在 Python 中？

Question

我有一個 train 文件夾，它包含 20 個類，每個 class 包含不同的 .png 和 .jpg 格式的圖像。 我已經在一個圖像上進行了測試，它工作正常，現在我想通過維護文件夾和 class 結構來將此過程應用於所有圖像。 在檢測到文本之后，我們有一個名為“final”的文件夾，它包含 20 個類，每個 class 內部包含一個檢測到的文本圖像。 檢測到的文本代碼已經可以正常工作

代碼

my_path = "Path"

files = glob.glob(my_path + '/**/*.jpg', recursive=True)
images = []

each_image = ''
for root, dirs, files in os.walk(my_path, topdown=False):
    for file in files:
        each_image = file
        print(root)
        print(dirs)

for file in files:

    image_path = file

    with open(image_path, 'rb') as image_file:
        content = image_file.read()

    image = vision.Image(content=content)
    response = client.text_detection(image=image)
    texts = response.text_annotations

    img_openCV = cv2.imread(image_path)
    h, w = img_openCV.shape[:2]
    print(h, w)

    for text in texts:
        box_w = abs(text.bounding_poly.vertices[2].x - text.bounding_poly.vertices[0].x)
        box_h = abs(text.bounding_poly.vertices[2].y - text.bounding_poly.vertices[0].y)

        if box_h * box_w > thres * h * w:
            continue
        if (text.bounding_poly.vertices[2].x - text.bounding_poly.vertices[0].x) > (w / 2):
            print('ebug width ')
            print(text.bounding_poly.vertices[2].x - text.bounding_poly.vertices[0].x)
        else:
            imagee = cv2.rectangle(img_openCV, (text.bounding_poly.vertices[0].x, text.bounding_poly.vertices[0].y),
                                   (text.bounding_poly.vertices[2].x, text.bounding_poly.vertices[2].y), (0, 255, 0),
                                   -1)
            print('height, width, color:', text.bounding_poly)
            cv2.imwrite(
                os.path.join("Path", each_image + str(total_images) + '.png'),
                imagee)
            total_images += 1

Answer 1

import os, io
import random
from turtle import color

from google.cloud import vision
import numpy as np
from PIL import Image, ImageDraw, ImageFont
import cv2
from text import draw_borders
import glob

os.environ["GOOGLE_APPLICATION_CREDENTIALS"] = "GoogleVisionAPI.json"
client = vision.ImageAnnotatorClient()

# file_name = '3017.jpg'
# image_folder = './img/'

total_images = 1
thres = 0.5

# Location with subdirectories
my_path = "/Path of whole folder which contains subclasses/"

files = glob.glob(my_path + '/**/*.jpg', recursive=True)
images = []

each_image = ''
# for root, dirs, files in os.walk(my_path, topdown=False):
#    for file in files:
#        each_image = file
#        print(root)
#        print(dirs)

for root, dirs, files in os.walk(my_path, topdown=False):
    for file in files:
        image_path = file

        with open(root + "/" + image_path, 'rb') as image_file:
            content = image_file.read()

            image = vision.Image(content=content)
            response = client.text_detection(image=image)
            texts = response.text_annotations

            img_openCV = cv2.imread(root + "/" + image_path)
            h, w = img_openCV.shape[:2]
            print(h, w)

            for text in texts:
                box_w = abs(text.bounding_poly.vertices[2].x - text.bounding_poly.vertices[0].x)
                box_h = abs(text.bounding_poly.vertices[2].y - text.bounding_poly.vertices[0].y)

                if box_h * box_w > thres * h * w:
                    continue
                if (text.bounding_poly.vertices[2].x - text.bounding_poly.vertices[0].x) > (w / 2):
                    print('ebug width ')
                    print(text.bounding_poly.vertices[2].x - text.bounding_poly.vertices[0].x)
                else:
                    imagee = cv2.rectangle(img_openCV,
                                           (text.bounding_poly.vertices[0].x, text.bounding_poly.vertices[0].y),
                                           (text.bounding_poly.vertices[2].x, text.bounding_poly.vertices[2].y),
                                           (0, 255, 0),
                                           -1)
                    print('height, width, color:', text.bounding_poly)
                    if root.__contains__("BC"):
                        cv2.imwrite(os.path.join("/Specific_class", file), imagee)
                    elif root.__contains__("BK"):
                        cv2.imwrite(os.path.join("/Specific_class", file), imagee)
                    elif root.__contains__("CC"):
                        cv2.imwrite(os.path.join("/Specific_class", file), imagee)
                    
             
            total_images += 1

# cv2.imshow('window_name', imagee)
# cv2.waitKey(0)

# closing all open windows
# cv2.destroyAllWindows()