opencv多目标追踪容器

之前做过一个多目标追踪的项目，尝试了一下opencv提供的追踪容器，个人感觉效果一般。

# coding:utf-8
# @Time : 14/12/2018 17:07
# @Author : SuRuiimport cv2
from functools import wrapsdef GetFps(func):@wraps(func)def wrapper(*args):timer = cv2.getTickCount()boxes = func(*args)fps = cv2.getTickFrequency() / (cv2.getTickCount() - timer)return boxes, fpsreturn wrapperclass MultipleTargetTracker:def __init__(self):self.multi_tracker = cv2.MultiTracker_create()self._method_name = None@propertydef method_name(self):return self._method_name@method_name.setterdef method_name(self, method):self._method_name = method@propertydef tracker_method(self):"""Choose one method from the follow list.'BOOSTING', 'CSRT', 'MIL', 'KCF', 'TLD', 'MEDIANFLOW', 'MOSSE', 'GOTURN':param method_name::return:"""# (major_ver, minor_ver, subminor_ver) = (cv2.__version__).split('.')tracker_types = ['BOOSTING', 'CSRT', 'MIL', 'KCF', 'TLD', 'MEDIANFLOW', 'MOSSE', 'GOTURN']tracker_generator = [cv2.TrackerBoosting_create, cv2.TrackerCSRT_create, cv2.TrackerMIL_create,cv2.TrackerKCF_create, cv2.TrackerTLD_create, cv2.TrackerMedianFlow_create,cv2.TrackerMOSSE_create, cv2.TrackerGOTURN_create]tracker_map = dict(zip(tracker_types, tracker_generator))tracker_method = tracker_map[self.method_name]return tracker_methoddef update_multi_tracker(self, image, boxes):self.multi_tracker = cv2.MultiTracker_create()# 必须对每个对象实例化一个追踪器才可以tracker_list = []for i in range(len(boxes)):tracker_list.append(self.tracker_method())x = boxes[i][0]y = boxes[i][1]w = boxes[i][2]h = boxes[i][3]#                             追踪器方法，     图片，  tuple（x, y, w, h）ok = self.multi_tracker.add(tracker_list[i], image, tuple([x, y, w, h]))# Update trackerok, bboxes = self.multi_tracker.update(image)print("-------- update ---------")print(bboxes)print("-------------------------")if not ok:raise NotImplementedError("Initial multiple-target tracker failed !")@GetFpsdef trace(self, frame):ok, boxes = self.multi_tracker.update(frame)if ok:return boxeselse:raise Exception("Tracking failure detected !")def view_result(tracker_method, frame, boxes, fps):for newbox in boxes:p1 = (int(newbox[0]), int(newbox[1]))p2 = (int(newbox[0] + newbox[2]), int(newbox[1] + newbox[3]))cv2.rectangle(frame, p1, p2, (200, 0, 0))# Display tracker type on frame              距离左上角的w和hcv2.putText(frame, tracker_method + " Tracker", (100, 20), cv2.FONT_HERSHEY_SIMPLEX, 0.75, (50, 170, 50), 2)# Display FPS on framecv2.putText(frame, "  FPS : " + str(int(fps)), (100, 50), cv2.FONT_HERSHEY_SIMPLEX, 0.75, (50, 170, 50), 2)return frameif __name__ == "__main__":mtt = MultipleTargetTracker()mtt.method_name = "KCF"

调用的demo.py，指定一个需要多个追踪对象的视频，认为标定4个roi信息即可开始执行追踪。

# coding:utf-8
import sys
import cv2 as cv
import numpy as np# if len(sys.argv) != 2:
#     print('Input video name is missing')
#     exit()print('Select 3 tracking targets')cv.namedWindow("tracking")
camera = cv.VideoCapture('../test_data/hiv30.mp4')
tracker = cv.MultiTracker_create()
init_once = False
second_once = False
ok, image = camera.read()
if not ok:print('Failed to read video')exit()bbox1 = cv.selectROI(image, False)
bbox2 = cv.selectROI(image, False)
bbox3 = cv.selectROI(image, False)
bbox4 = cv.selectROI(image, False)print(type(bbox1))
print(bbox1)
newboxs = []
while camera.isOpened():ok, image = camera.read()if not ok:print('no image to read')breakif not init_once:ok = tracker.add(cv.TrackerKCF_create(), image, bbox1)ok = tracker.add(cv.TrackerKCF_create(), image, bbox2)ok = tracker.add(cv.TrackerKCF_create(), image, bbox3)init_once = Trueok, boxes = tracker.update(image)print(ok, boxes)if not second_once:ok = tracker.add(cv.TrackerKCF_create(), image, bbox4)second_once = Trueok, boxes = tracker.update(image)if ok:newboxs = boxesif not ok:del trackerprint('tracker 消失')tracker = cv.MultiTracker_create()for box in newboxs:box = tuple(box)ok = tracker.add(cv.TrackerKCF_create(), image, box)for newbox in boxes:p1 = (int(newbox[0]), int(newbox[1]))p2 = (int(newbox[0] + newbox[2]), int(newbox[1] + newbox[3]))cv.rectangle(image, p1, p2, (200, 0, 0))cv.imshow('tracking', image)k = cv.waitKey(1)if k == 27: break  # esc pressed

opencv多目标追踪容器相关推荐

OpenCV图像处理——目标追踪
总目录图像处理总目录←点击这里二十四.目标追踪 24.1.多目标(手动检测)追踪 24.1.1.原理目标检测:运行之后按下s,通过鼠标对某个目标进行检测,然后点击空格或者回车目标追踪:open ...
OpenCV camshift目标追踪
camshift算法是对meanshift算法的改进,首先应用meanshift,一旦meanshift收敛,它就会更新窗口的大小,还计算最佳拟合椭圆的方向,从而根据目标的位置和大小更新搜索窗口. c ...
OpenCV meanshift目标追踪
meanshift原理:一个迭代的步骤,即先算出当前点的偏移均值,移动该点到其偏移均值,然后以此为新的起始点,继续移动,直到满足一定的条件结束. Meanshift算法:简单,迭代次数少,但无法解决目 ...
【Opencv】目标追踪——高斯混合模型分离算法(MOG)
文章目录 1 环境 2 效果 3 原理 4 代码 1 环境 Python 3.8.8 PyCharm 2021 opencv-python 2 效果 3 原理视频图像中的目标检测与跟踪,是计算机 ...
python—opencv 实时视频目标追踪
功能: 1.获取摄像头,实时显示 2.鼠标获取第一帧中的目标roi区域 3.在视频中实时对目标进行追踪. 4.两种目标追踪的方式:'meanshift','camshift' 5.保存视频 impor ...
用opencv实现目标追踪的学习笔记——camshift
小白的学习笔记--opencv camshift -基础:零c++基础,零opencv基础,简单C语言基础,略知数字图像处理知识 -工具:VS2015+opencv 2.4.13 -sample: E ...
python机器识别追踪_多目标追踪器：用OpenCV实现多目标追踪（C++/Python)
原标题:多目标追踪器:用OpenCV实现多目标追踪(C++/Python) MultiTracker : Multiple Object Tracking using OpenCV (C++/Pyth ...
基于opencv的BackgroundSubtractorMOG2目标追踪
基于opencv的BackgroundSubtractorMOG2目标追踪 BackgroundSubtractorMOG2是以高斯混合模型为基础的背景/前景分割算法.它是以2004年和2006年Z. ...

opencv多目标追踪容器

opencv多目标追踪容器

opencv多目标追踪容器相关推荐

最新文章

热门文章