交通标志图像处理，已知感兴趣区域.csv文件，如何循环使图片与.csv中每一行对应，从而进行mask处理？

Question

已知感兴趣区域.csv文件，想通过循环使图片与.csv中每一行对应，从而进行mask处理，将不感兴趣区域变成黑色，但是好像循环的不对，恳请大家帮助

def Image_Resize(dir):   
    for parent,dirnames,filenames in os.walk(dir):      
        for filename in filenames:

            save_path = os.path.join('D:/Testing_mask/00000', filename)
            csv_data = pd.read_csv('D:/BLS_Test_Images_jpg/00000/GT-00000.csv')
            csv_data_array = np.array(csv_data)
            for i in range(csv_data_array.shape[0]):
                csv_data_list = np.array(csv_data)[i, :].tolist()[0].split(";")
                sample_dir = os.path.join(save_path, csv_data_list[0])
                img = cv2.imread(sample_dir)
                mask = np.zeros(img.shape[:2], np.uint8)
                mask[int(csv_data_list[3]):int(csv_data_list[5]), int(csv_data_list[4]):int(csv_data_list[6])] = 255

                cv2.imwrite(sample_dir, masked_img)


Image_Resize('D:/Testing_mask/00000')

Answer 1

以下回答参考 皆我百晓生、券券喵儿 等免费微信小程序相关内容作答，并由本人整理回复。

import numpy as np
from PIL import Image

# Load the CSV file and extract the image paths and GT data
csv_file = 'your_csv_file.csv'
data = pd.read_csv(csv_file)

# Define the directory where your images are stored
image_directory = 'your_image_directory'

# Iterate through each row of the CSV file
for index, row in data.iterrows():
    # Extract the image path and GT data from the current row
    image_path = row['Image Path']
    gt_data = row['GT Data'].split(';')

    # Open the image using Pillow library
    img = Image.open(image_path)

    # Create a new image with transparent background
    new_img = Image.new("RGBA", img.size, (0, 0, 0, 0))

    # Paste the original image onto the new image
    new_img.paste(img, (0, 0), img.convert("RGBA"))

    # Get the width and height of the image
    w, h = img.size

    # Calculate the center of the image
    center_x = w / 2
    center_y = h / 2

    # Set the ROI to be the area specified by the GT data
    roi = (center_x - int(gt_data[1]), center_y - int(gt_data[2]),
           center_x + int(gt_data[3]) - int(gt_data[1]),
           center_y + int(gt_data[4]) - int(gt_data[2]))

    # Crop the image to the specified ROI
    cropped_img = new_img.crop(roi)

    # Save the cropped image with transparency
    cropped_img.save(f'{image_directory}/{row["ID"]}.png', "PNG")

print("Done!")

注意：这个脚本使用了Pillow库来操作图像。如果你还没有安装Pillow，请先在命令行中运行以下命令：

pip install pillow

然后替换your_csv_file.csv和your_image_directory为你的实际CSV文件路径和图像目录路径。