背景简述
我们有一批标注项目要转可视化,因为之前没有做过,然后网上随意找了一段代码测试完美(并没有)搞定,开始疯狂标注,当真正要转的时候傻眼了,因为测试的时候用的是英文标签,实际标注的是中文标签,结果都是一大堆??????,
结果瞬间让我满脑袋??????,赶紧找资料解决,各种方法试了个遍,网上大多数都是用cv2+matplotlib实现的计算和渲染,所以解决的主要思想就是集中在各种显示的设置matplotlib字体,然并卵;最后找到一种另辟蹊径的办法使用PIL+cv2实现,最后完美解决,
贴上解决代码:
import cv2
import os
import numpy as np
from PIL import Image, ImageDraw, ImageFont
import xml.etree.ElementTree as ET
data_path = 'E:\\test\\tianjingulou'
imgs_path = os.path.join(data_path, "img")
anns_path = os.path.join(data_path, "xml")
result_path = os.path.join(data_path)
img_names = set([os.path.splitext(i)[0] for i in os.listdir(imgs_path)])
ann_names = set([os.path.splitext(i)[0] for i in os.listdir(anns_path)])
img_names = list(img_names)
ann_names = list(ann_names)
for i in range(len(img_names)):
img_path = os.path.join(imgs_path, img_names[i] + ".jpg")
img_bgr = cv2.imread(img_path)
xml_path = os.path.join(anns_path, ann_names[i] + ".xml")
xml_inf = open(xml_path, encoding='utf-8')
tree = ET.parse(xml_inf)
root = tree.getroot()
bbox_color = (0, 129, 255)
bbox_thickness = 2
# 把rgb转成16进制'#0081FF'
bbox_color_str = "#{:02x}{:02x}{:02x}".format(*bbox_color)
# 把rgb转成bgr再转16进制'#FF8100'
# bbox_color_rgb = bbox_color[::-1]
# bbox_color_str = "#{:02x}{:02x}{:02x}".format(*bbox_color_rgb)
bbox_labelstr = {
'font_size': 16,
'font_thickness': 2,
'offset_x': 0,
'offset_y': -20,
}
# 创建一个空白图像
img_pil = Image.fromarray(cv2.cvtColor(img_bgr, cv2.COLOR_BGR2RGB))
draw = ImageDraw.Draw(img_pil)
# 设置字体 SimHei.ttf黑体,msyh.ttf微软雅黑
# 打开命令行窗口或者Anaconda Prompt,输入python,进入python解释器窗口,
# 输入import matplotlib;引入可视化库;
# 然后输入print(matplotlib.matplotlib_fname())打印出当前库所在位置;
# 进入到上面打印出的路径下字体目录:mpl-data\\fonts\\ttf,下载中文字体放进去
font_path = "D:\\ProgramData\\anaconda3\\Lib\\site-packages\\matplotlib\\mpl-data\\fonts\\ttf\\msyh.ttf" # 请替换为实际路径
font = ImageFont.truetype(font_path, bbox_labelstr['font_size'])
# 画框和文字
for obj in root.iter('object'):
bbox_label = obj.find('name').text
bbox_top_left_x = int(obj.find('bndbox').find('xmin').text)
bbox_top_left_y = int(obj.find('bndbox').find('ymin').text)
bbox_bottom_right_x = int(obj.find('bndbox').find('xmax').text)
bbox_bottom_right_y = int(obj.find('bndbox').find('ymax').text)
draw.rectangle([(bbox_top_left_x, bbox_top_left_y), (bbox_bottom_right_x, bbox_bottom_right_y)],
outline=bbox_color, width=bbox_thickness)
draw.text((bbox_top_left_x + bbox_labelstr['offset_x'], bbox_top_left_y + bbox_labelstr['offset_y']),
bbox_label, font=font, fill=bbox_color_str)
img_bgr = cv2.cvtColor(np.array(img_pil), cv2.COLOR_RGB2BGR)
# 保存图像
cv2.imwrite(result_path + "\\{}.jpg".format(img_names[i]), img_bgr)
下面是matplotlib+cv2版代码
# 数据集可视化
import cv2
import os
import matplotlib.pyplot as plt
import xml.etree.ElementTree as ET
# 设置 Matplotlib 使用的字体为黑体
plt.rcParams['font.sans-serif'] = ['SimHei']
plt.rcParams['axes.unicode_minus'] = False
imgs_path = 'E:\\test\\tianjingulou\\img'
anns_path = 'E:\\test\\tianjingulou\\xml'
img_names = set([os.path.splitext(i)[0] for i in os.listdir(imgs_path)])
ann_names = set([os.path.splitext(i)[0] for i in os.listdir(anns_path)])
img_names = list(img_names)
ann_names = list(ann_names)
for i in range(len(img_names)):
img_path = os.path.join(imgs_path, img_names[i] + ".jpg")
img_bgr = cv2.imread(img_path)
xml_path = os.path.join(anns_path, ann_names[i] + ".xml")
xml_inf = open(xml_path, encoding='utf-8')
tree = ET.parse(xml_inf)
root = tree.getroot()
# 框可视化配置
bbox_color = (255, 129, 0) # 框的颜色
bbox_thickness = 2 # 框的线宽
# 框类别文字
bbox_labelstr = {
'font_size': 1, # 字体大小
'font_thickness': 2, # 字体粗细
'offset_x': 0, # X 方向,文字偏移距离,向右为正
'offset_y': -10, # Y 方向,文字偏移距离,向下为正
}
# 画框
for obj in root.iter('object'): # 一个object代表一个标注物体
# 框的类别
bbox_label = obj.find('name').text
# 框的两点坐标
# 左上角坐标
bbox_top_left_x = int(obj.find('bndbox').find('xmin').text)
bbox_top_left_y = int(obj.find('bndbox').find('ymin').text)
# 右下角坐标
bbox_bottom_right_x = int(obj.find('bndbox').find('xmax').text)
bbox_bottom_right_y = int(obj.find('bndbox').find('ymax').text)
# 画矩形:画框
img_bgr = cv2.rectangle(img_bgr, (bbox_top_left_x, bbox_top_left_y), (bbox_bottom_right_x, bbox_bottom_right_y),
bbox_color, bbox_thickness)
# 写框类别文字:图片,文字字符串,文字左上角坐标,字体,字体大小,颜色,字体粗细
img_bgr = cv2.putText(img_bgr, bbox_label, (
bbox_top_left_x + bbox_labelstr['offset_x'],
bbox_top_left_y + bbox_labelstr['offset_y']),
cv2.FONT_HERSHEY_SIMPLEX, bbox_labelstr['font_size'], bbox_color,
bbox_labelstr['font_thickness'])
cv2.imwrite("E:\\test\\tianjingulou\\{}.jpg".format(img_names[i]), img_bgr)
写在最后,matplotlib的方式应该也有解决的办法,也可能是我的环境问题,提供这两种方式大家各取所需,下面这种方式是我从一位博主那里拷贝来稍加改动的,但是我找不到出处了,如有侵权请联系我删除。
----------------------------------------------华丽分割-------------------------------------------------
追加一种类似的写法,这个是宋体,字体可以酌情替换,亲测可用
import cv2
import os
import matplotlib.pyplot as plt
import xml.etree.ElementTree as ET
import numpy as np
# 导入 PIL 库
import PIL.Image
import PIL.ImageDraw
import PIL.ImageFont
data_path = os.path.join("E:\\test\\tianjingulou")
imgs_path = os.path.join(data_path, "img")
anns_path = os.path.join(data_path, "xml")
# 获取图像名称和标注名称
img_names = set(os.path.splitext(i)[0] for i in os.listdir(imgs_path))
ann_names = set(os.path.splitext(i)[0] for i in os.listdir(anns_path))
img_names = list(img_names)
ann_names = list(ann_names)
# 遍历所有图像
for i, img_name in enumerate(img_names):
# 读取图像
img_bgr = cv2.imread(os.path.join(imgs_path, img_name + ".jpg"))
# 读取标注
xml_path = os.path.join(anns_path, img_name + ".xml")
xml_inf = open(xml_path, encoding='utf-8')
tree = ET.parse(xml_inf)
root = tree.getroot()
# 画框
for obj in root.iter('object'):
# 获取框的类别
bbox_label = obj.find('name').text
# 获取框的两点坐标
bbox_top_left_x = int(obj.find('bndbox').find('xmin').text)
bbox_top_left_y = int(obj.find('bndbox').find('ymin').text)
bbox_bottom_right_x = int(obj.find('bndbox').find('xmax').text)
bbox_bottom_right_y = int(obj.find('bndbox').find('ymax').text)
# 画矩形
img_bgr = cv2.rectangle(img_bgr, (bbox_top_left_x, bbox_top_left_y), (bbox_bottom_right_x, bbox_bottom_right_y),
(255, 129, 0), 2)
# 写框类别文字
# 转换为 PIL 图像
img_pil = PIL.Image.fromarray(img_bgr)
# 使用 PIL 绘制文本
font = PIL.ImageFont.truetype("simsun.ttc", 16)
draw = PIL.ImageDraw.Draw(img_pil)
draw.text((bbox_top_left_x, bbox_top_left_y - 18), bbox_label, font=font, fill=(255, 129, 0))
# 直接使用 PIL 图像
img_bgr = np.array(img_pil)
# 保存图像
cv2.imwrite(data_path + "\\{}.jpg".format(img_name), img_bgr)