从子文件夹中读取所有图像，检测文本并在 Python 中使用原始名称和文件夹保存它们？

Question

我有一个 train 文件夹，它包含 20 个类，每个类包含 .png 和 .jpg 格式的不同图像。我已经在一个图像上进行了测试，它工作正常，现在我想通过维护文件夹和类结构将这个过程应用于所有图像。在检测到文本之后，我们有一个名为“最终”名称的文件夹，它包含 20 个类，每个类内部都包含一个检测到的文本图像。检测到的文本代码已经可以正常工作了

代码

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()