做目标检测的首要前提是对自己的数据了如指掌，有时候使用其他人的数据集，第一时间可能不知道其他人的打标签标准是什么，那么我们可以可视出来目标框，把标签文件里的坐标绘制到原图上，下面程序只需要给定图片、标签和保存位置，就可以直接使用。
# -*- coding: utf-8 -*-
from __future__ import division
import os
import xml.dom.minidom
import cv2
import sys
import numpy as np
# from imp import reload
# reload(sys)
def read_xml(ImgPath, AnnoPath, Savepath):
imagelist = os.listdir(AnnoPath)
for image in imagelist:
image_pre, ext = os.path.splitext(image)
# imgfile = +'/'+ image_pre+ '.JPG'
imgfile = os.path.join(ImgPath,image_pre+ '.jpg')
# xmlfile = AnnoPath +'/'+ image_pre+ '.xml'
xmlfile = os.path.join(AnnoPath, image_pre + '.xml')
print(imgfile)
print(xmlfile)
# im = cv2.imread(imgfile)
im = cv2.imdecode(np.fromfile(imgfile,dtype=np.uint8),cv2.IMREAD_UNCHANGED)#imdecode()读取图像数据并转换成图片格式
#fromfile()读数据时需要用户指定元素类型，并对数组的形状进行适当的修改，cv2.IMREAD_UNCHANGED加载图像
DomTree = xml.dom.minidom.parse(xmlfile)#读取xml文件中的值
annotation = DomTree.documentElement #documentElement 属性可返回文档的根节点。
filenamelist = annotation.getElementsByTagName('filename')#getElementById()可以访问Documnent中的某一特定元素，顾名思义，就是通过ID来取得元素，所以只能访问设置了ID的元素。
filename = filenamelist[0].childNodes[0].data
objectlist = annotation.getElementsByTagName('object')
i = 1
for objects in objectlist:
namelist = objects.getElementsByTagName('name')
objectname = namelist[0].childNodes[0].data #通过xml文件给图像加目标框
bndbox = objects.getElementsByTagName('bndbox')
for box in bndbox:
try:
x1_list = box.getElementsByTagName('xmin')
x1 = int(x1_list[0].childNodes[0].data)
y1_list = box.getElementsByTagName('ymin')
y1 = int(y1_list[0].childNodes[0].data)
x2_list = box.getElementsByTagName('xmax')
x2 = int(x2_list[0].childNodes[0].data)
y2_list = box.getElementsByTagName('ymax')
y2 = int(y2_list[0].childNodes[0].data)
minX = x1
minY = y1
maxX = x2
maxY = y2
if(i % 3 == 0):
color = (128,0,0)
elif (i % 3 == 1):
color = (153, 51, 0)
elif (i % 3 == 2):
color = (255, 204, 0)
elif (i % 3 == 3):
color = (0, 51, 0)
elif (i % 9 == 4):
color = (51, 204, 204)
elif (i % 9 == 5):
color = (128, 0, 128)
elif (i % 9 == 6):
color = (0, 255, 255)
elif (i % 9 == 7):
color = (60, 179, 113)
elif (i % 9 == 8):
color = (255, 127, 80)
elif (i % 9 == 9):
color = (0, 255, 0)
cv2.rectangle(im,(minX,minY),(maxX,maxY),color,8)
if not os.path.exists(Savepath):
os.makedirs(Savepath)
path = os.path.join(Savepath, image_pre + '.jpg')
font = cv2.FONT_HERSHEY_SIMPLEX
cv2.putText(im, objectname, (minX,minY - 7), font, 0.7, (0, 0, 255), 2)
cv2.imencode(".jpg",im)[1].tofile(path)
i += 1
except Exception as e:
print(e)
if __name__ == "__main__":
img_path = r'F:\VOC2007_test\JPEGImages/'
xml_path = r'F:\VOC2007_test\Annotations/'
save_path = r'F:\VOC2007_test\save/'
read_xml(img_path, xml_path,save_path)
我们还可以通过修改算法，让相同类别的标签框绘制成同样的颜色，代码如下：
# -*- coding: utf-8 -*-
from __future__ import division
import os
import xml.dom.minidom
import cv2
import sys
import numpy as np
def read_xml(ImgPath, AnnoPath, Savepath):
imagelist = os.listdir(AnnoPath)
for image in imagelist:
image_pre, ext = os.path.splitext(image)
# imgfile = +'/'+ image_pre+ '.JPG'
imgfile = os.path.join(ImgPath,image_pre+ '.jpg')
# xmlfile = AnnoPath +'/'+ image_pre+ '.xml'
xmlfile = os.path.join(AnnoPath, image_pre + '.xml')
print(imgfile)
print(xmlfile)
# im = cv2.imread(imgfile)
im = cv2.imdecode(np.fromfile(imgfile,dtype=np.uint8),cv2.IMREAD_UNCHANGED)#imdecode()读取图像数据并转换成图片格式
#fromfile()读数据时需要用户指定元素类型，并对数组的形状进行适当的修改，cv2.IMREAD_UNCHANGED加载图像
DomTree = xml.dom.minidom.parse(xmlfile)#读取xml文件中的值
annotation = DomTree.documentElement #documentElement 属性可返回文档的根节点。
filenamelist = annotation.getElementsByTagName('filename')#getElementById()可以访问Documnent中的某一特定元素，顾名思义，就是通过ID来取得元素，所以只能访问设置了ID的元素。
filename = filenamelist[0].childNodes[0].data
objectlist = annotation.getElementsByTagName('object')
i = 1
for objects in objectlist:
namelist = objects.getElementsByTagName('name')
objectname = namelist[0].childNodes[0].data #通过xml文件给图像加目标框
bndbox = objects.getElementsByTagName('bndbox')
for box in bndbox:
try:
x1_list = box.getElementsByTagName('xmin')
x1 = int(x1_list[0].childNodes[0].data)
y1_list = box.getElementsByTagName('ymin')
y1 = int(y1_list[0].childNodes[0].data)
x2_list = box.getElementsByTagName('xmax')
x2 = int(x2_list[0].childNodes[0].data)
y2_list = box.getElementsByTagName('ymax')
y2 = int(y2_list[0].childNodes[0].data)
minX = x1
minY = y1
maxX = x2
maxY = y2
if objectname=="类别1":
color = (255, 0, 80)
if objectname=="类别2":
color = (255,0,255)
if objectname=="类别3":
color = (51, 204, 204)
if objectname=="类别4":
color = (255, 204, 0)
cv2.rectangle(im,(minX,minY),(maxX,maxY),color,8)
if not os.path.exists(Savepath):
os.makedirs(Savepath)
path = os.path.join(Savepath, image_pre + '.jpg')
font = cv2.FONT_HERSHEY_SIMPLEX
cv2.putText(im, objectname, (minX,minY - 7), font, 0.7, (0, 0, 255), 2)
#print(path)
# cv2.imwrite(path, im)
cv2.imencode(".jpg",im)[1].tofile(path)
i += 1
except Exception as e:
print(e)
if __name__ == "__main__":
img_path = r'F:\VOC2007_test\JPEGImages/'
xml_path = r'F:\VOC2007_test\Annotations/'
save_path = r'F:\VOC2007_test\save/'
read_xml(img_path, xml_path,save_path)
版权声明：本文为a_1234567890_原创文章，遵循 CC 4.0 BY-SA 版权协议，转载请附上原文出处链接和本声明。