Python使用OPENCV的目标跟踪算法实现自动视频标注效果

先上效果

 1.首先，要使用opencv的目标跟踪算法，必须要有opencv环境

 使用：opencv==4.4.0 和 opencv-contrib-python==4.4.0.46，lxml   这三个环境包。

也可以使用以下方法进行下载 ： 

pip install opencv-python==4.4.0
pip install opencv-contrib-python==4.4.0.4

pip install lxml

2.使用方法：

        （1）：英文状态下的 “s” 是进行标注

        （2）：使用小键盘 1-9 按下对应的标签序号，标签序号和标签可自定义（需要提前定义）

       （3）：对目标进行绘制

       （4）：按空格键继续

        重复进行 （1）（2）（3）（4）步骤，可实现多个目标的跟踪绘制

        英文状态下的 “r” 是所有清除绘制

         英文状态下的 “q” 是退出

          当被跟踪目标丢失时，自动清除所有绘制

import cv2
import os
import time
from lxml import etree
 
#视频路径
Vs = cv2.VideoCapture(\'peaple.avi\')
#自定义标签
Label = {1:\"people\",2:\"car\",3:\"Camera\"}
#图片保存路径 ，一定使用要用绝对路径！！
imgpath = r\"C:\\Users\\BGT\\Desktop\\opencv\\img\"
#xml保存路径 ，一定使用要用绝对路径！！
xmlpath = r\"C:\\Users\\BGT\\Desktop\\opencv\\xml\"
#设置视频缩放
cv2.namedWindow(\"frame\", 0)
#设置视频宽高
cv2.resizeWindow(\"frame\", 618, 416)
 
#定义生成xml类
class Gen_Annotations:
    def __init__(self, json_info):
        self.root = etree.Element(\"annotation\")
 
        child1 = etree.SubElement(self.root, \"folder\")
        child1.text = str(json_info[\"pic_dirname\"])
 
        child2 = etree.SubElement(self.root, \"filename\")
        child2.text = str(json_info[\"filename\"])
 
        child3 = etree.SubElement(self.root, \"path\")
        child3.text = str(json_info[\"pic_path\"])
 
        child4 = etree.SubElement(self.root, \"source\")
 
        child5 = etree.SubElement(child4, \"database\")
        child5.text = \"My name is BGT\"
 
    def set_size(self, witdh, height, channel):
        size = etree.SubElement(self.root, \"size\")
        widthn = etree.SubElement(size, \"width\")
        widthn.text = str(witdh)
        heightn = etree.SubElement(size, \"height\")
        heightn.text = str(height)
        channeln = etree.SubElement(size, \"depth\")
        channeln.text = str(channel)
        segmented = etree.SubElement(self.root, \"segmented\")
        segmented.text = \"0\"
 
    def savefile(self, filename):
        tree = etree.ElementTree(self.root)
        tree.write(filename, pretty_print=True, xml_declaration=False, encoding=\'utf-8\')
 
    def add_pic_attr(self, label, x0, y0, x1, y1):
        object = etree.SubElement(self.root, \"object\")
        namen = etree.SubElement(object, \"name\")
        namen.text = label
        pose = etree.SubElement(object, \"pose\")
        pose.text = \"Unspecified\"
        truncated = etree.SubElement(object, \"truncated\")
        truncated.text = \"0\"
        difficult = etree.SubElement(object, \"difficult\")
        difficult.text = \"0\"
        bndbox = etree.SubElement(object, \"bndbox\")
        xminn = etree.SubElement(bndbox, \"xmin\")
        xminn.text = str(x0)
        yminn = etree.SubElement(bndbox, \"ymin\")
        yminn.text = str(y0)
        xmaxn = etree.SubElement(bndbox, \"xmax\")
        xmaxn.text = str(x1)
        ymaxn = etree.SubElement(bndbox, \"ymax\")
        ymaxn.text = str(y1)
 
 #定义生成xml的方法
def voc_opencv_xml(a,b,c,d,e,f,boxes,Label,Label_a,save=\"1.xml\"):
    json_info = {}
    json_info[\"pic_dirname\"] = a
    json_info[\"pic_path\"] = b
    json_info[\"filename\"] = c
    anno = Gen_Annotations(json_info)
 
    anno.set_size(d, e, f)
 
    for box in range(len(boxes)):
        x,y,w,h = [int(v) for v in boxes[box]]
        anno.add_pic_attr(Label[Label_a[box]],x,y,x+w,y+h)
    anno.savefile(save)
  
if __name__ == \'__main__\':
    Label_a = []
    contents = os.path.split(imgpath)[1]
    trackers = cv2.MultiTracker_create()
    while True:
        Filename_jpg = str(time.time()).split(\".\")[0] + \"_\" + str(time.time()).split(\".\")[1] + \".jpg\"
        Filename_xml = str(time.time()).split(\".\")[0] + \"_\" + str(time.time()).split(\".\")[1] + \".xml\"
 
        path_Filename_jpg = os.path.join(imgpath,Filename_jpg)
        path_Filename_xml = os.path.join(xmlpath,Filename_xml)
 
        ret,frame = Vs.read()
        if not ret:
            break
 
        success,boxes = trackers.update(frame)
        if len(boxes)>0:
            cv2.imwrite(path_Filename_jpg, frame)
            judge = True
        else:
            judge = False
 
         if success==False:
            print(\"目标丢失\")
            trackers = cv2.MultiTracker_create()
            Label_a = []
            judge = False
        if judge:
            voc_opencv_xml(contents,Filename_jpg,path_Filename_jpg,frame.shape[1],frame.shape[0],frame.shape[2],boxes,Label,Label_a,path_Filename_xml)
        if judge:
            for box in range(len(boxes)):
                x,y,w,h = [int(v) for v in boxes[box]]
                cv2.putText(frame, Label[Label_a[box]], (x, y), cv2.FONT_HERSHEY_COMPLEX, 1, (255, 255, 255), 1)
                cv2.rectangle(frame,(x,y),(x+w,y+h),(255,0,0),2)
  
        cv2.imshow(\'frame\',frame)
 
        var = cv2.waitKey(30)
 
        if var == ord(\'s\'):
            imgzi = cv2.putText(frame, str(Label), (50, 50), cv2.FONT_HERSHEY_TRIPLEX, 1, (0, 255, 0), 2)
            cv2.imshow(\'frame\', frame)
            var = cv2.waitKey(0)
            if var-48<len(Label) or var-48<=len(Label):
                Label_a.append(int(var-48))
            box = cv2.selectROI(\"frame\", frame, fromCenter=False,showCrosshair=True)
            tracker = cv2.TrackerCSRT_create()
            trackers.add(tracker,frame,box)
        elif var == ord(\"r\"):
            trackers = cv2.MultiTracker_create()
            Label_a = []
        elif var == ord(\'q\'): #退出
            break
 
    Vs.release()
    cv2.destroyAllWindows()
 

3.得到xml和img数据是VOC格式，img和xml文件以时间戳进行命名。防止同名覆盖。

4.最后使用 labelImg软件  对获取到的img和xml进行最后的检查和微调