学习记录1——xml格式数据集标签可视化

将xml格式的数据集标签可视化，在原图的相应位置绘制矩形框和目标类别

原创代码，只需使用PIL库和python中自带的xml.dom.minidom包（本人技术水平有限，使用的方法比较简单粗暴）修改文件路径即可使用。

import os
from xml.dom import minidom  # python自带的xml.dom.minidom包
from PIL import ImageDraw, ImageFont, Image  # PIL绘制图形、文本

# 定义函数draw_rectangle画矩形框**********************************
def draw_rectangle(name):
    doc = minidom.parse("D:/test/" + name)  # 打开xml文件，注意文件名格式
    imageName = name.replace(".xml", ".jpg")  # 替换文件名后缀
    image = Image.open("D:/test/" + imageName)  # 打开相应的jpg文件

    objects = doc.getElementsByTagName("object")  # 定义objects
    for object in objects:  # 遍历标签中所有的物体
        name = object.getElementsByTagName("name")[0]
        xmin = object.getElementsByTagName("xmin")[0]
        ymin = object.getElementsByTagName("ymin")[0]
        xmax = object.getElementsByTagName("xmax")[0]
        ymax = object.getElementsByTagName("ymax")[0]
        # type(xmin) = xml.dom.minidom.Element

        x_min = int("".join([str(x1) for x1 in xmin.firstChild.data]))
        y_min = int("".join([str(y1) for y1 in ymin.firstChild.data]))
        x_max = int("".join([str(x2) for x2 in xmax.firstChild.data]))
        y_max = int("".join([str(y2) for y2 in ymax.firstChild.data]))
        # type(xmin.firstChild.data) = str      type(x_min) = int
        # str转int: int("".join([str(x) for x in str]))

        draw = ImageDraw.Draw(image)  # 加载画图命令
        if name.firstChild.data == "truck":
            draw.rectangle([x_min, y_min, x_max, y_max], outline=color[0], width=3)  
                          # (坐标，颜色，线宽)
            draw.text((x_min, y_min - 30), "truck", fill=color[0], font=ttfont)  
                     # (坐标，文本，颜色，字体+大小)
        if name.firstChild.data == "car":
            draw.rectangle([x_min, y_min, x_max, y_max], outline=color[1], width=3)
            draw.text((x_min, y_min - 30), "car", fill=color[1], font=ttfont)
        if name.firstChild.data == "dangertruck":
            draw.rectangle([x_min, y_min, x_max, y_max], outline=color[2], width=3)
            draw.text((x_min, y_min - 30), "dangertruck", fill=color[2], font=ttfont)
        if name.firstChild.data == "special truck":
            draw.rectangle([x_min, y_min, x_max, y_max], outline=color[3], width=3)
            draw.text((x_min, y_min - 30), "special truck", fill=color[3], font=ttfont)
        if name.firstChild.data == "bus":
            draw.rectangle([x_min, y_min, x_max, y_max], outline=color[4], width=3)
            draw.text((x_min, y_min - 30), "bus", fill=color[4], font=ttfont)
        if name.firstChild.data == "person":
            draw.rectangle([x_min, y_min, x_max, y_max], outline=color[5], width=3)
            draw.text((x_min, y_min - 30), "person", fill=color[5], font=ttfont)
        if name.firstChild.data == "motorcycle":
            draw.rectangle([x_min, y_min, x_max, y_max], outline=color[6], width=3)
            draw.text((x_min, y_min - 30), "motorcycle", fill=color[6], font=ttfont)
        if name.firstChild.data == "bicycle":
            draw.rectangle([x_min, y_min, x_max, y_max], outline=color[7], width=3)
            draw.text((x_min, y_min - 30), "bicycle", fill=color[7], font=ttfont)
        if name.firstChild.data == "animal":
            draw.rectangle([x_min, y_min, x_max, y_max], outline=color[8], width=3)
            draw.text((x_min, y_min - 30), "animal", fill=color[8], font=ttfont)

    image.save("D:/Image/" + imageName)  # 保存图像，C盘根目录无权限
# 定义函数draw_rectangle画矩形框**********************************

file_dir = "D:/test"  # path
file_name = os.listdir(file_dir)  # 遍历文件获取全部文件名
color = [(102, 128, 102), (255, 0, 0), (0, 255, 0), (0, 0, 255), (255, 255, 0), (255, 0, 255), (0, 255, 255),
             (0, 128, 0), (0, 0, 0)]  # 控制矩形框颜色
ttfont = ImageFont.truetype("C:\Windows\Fonts\Arial.ttf", 30)  # 控制文本字体（电脑自带字体库）与大小
for name in file_name:  # 循环语句（注意:）
    if name.find(".xml") >= 0:  # 仅读取xml文件
       draw_rectangle(name)

效果如下：