基于yolov8，制作停车位计数器(附源码)

大家好，YOLO(You Only Look Once) 是由Joseph Redmon和Ali开发的一种对象检测和图像分割模型。

YOLO的第一个版本于2015年发布，由于其高速度和准确性，瞬间得到了广大AI爱好者的喜爱。

Ultralytics YOLOv8则是一款前沿、最先进(SOTA)的模型，基于先前YOLO版本的成功，引入了新功能和改进，进一步提升性能和灵活性。

YOLOv8设计快速、准确且易于使用，使其成为各种物体检测与跟踪、实例分割、图像分类和姿态估计任务的绝佳选择。

项目地址：

https://github.com/ultralytics/ultralytics

其中官方提供了示例，通过Python代码即可实现YOLOv8对象检测算法模型，使用预训练模型来检测我们的目标。而且对电脑需求也不高，CPU就能运行代码。

今天我就给大家介绍三个使用YOLOv8制作的计数器，非常实用。使用到Python版本以及相关Python库。

Python 3.9.7  
ultralytics 8.0.178  
opencv-contrib-python 4.8.1.78  
opencv-python 4.8.0.74  
cvzone 1.5.6

技术交流群

前沿技术资讯、算法交流、求职内推、算法竞赛、面试交流(校招、社招、实习)等、与 10000+来自港科大、北大、清华、中科院、CMU、腾讯、百度等名校名企开发者互动交流~

建了技术交流群&星球！想要本文源码、进交流群的同学，可以直接加微信号：mlc2060。加的时候备注一下：研究方向 +学校/公司，即可。然后就可以拉你进群了。

方式①、添加微信号：mlc2060，备注：技术交流
方式②、微信搜索公众号：机器学习社区，后台回复：技术交流

/ 01 /

客流检测器

使用OpenCV检测顾客，并且设定客人进出区域，实现实时计算进出顾客的数量。

客流量统计对于零售行业来说是非常重要的。

统计每天的进出店人数、过店人数以及人均驻留时间等。

依据这些数据，经营者可以对店铺的经营策略进行调整，实现店铺的经营效益最大化。

接下来就来看一下客流计数器的检测代码吧！

import cv2  
import numpy as np  
from tracker import \*  
import cvzone  
import time  bg\_subtractor = cv2.createBackgroundSubtractorMOG2(history=200, varThreshold=140)  # 打开视频  
video\_capture = cv2.VideoCapture(r"store.mp4")  def RGB(event, x, y, flags, param):  if event == cv2.EVENT\_MOUSEMOVE:  point = \[x, y\]  print(point)  cv2.namedWindow('RGB')  
cv2.setMouseCallback('RGB', RGB)  
tracker = Tracker()  area1 = \[(213, 165), (200, 189), (693, 373), (697, 341)\]  
area2 = \[(195, 199), (186, 213), (683, 404), (689, 388)\]  
er = {}  
counter1 = \[\]  
ex = {}  
counter2 = \[\]  
while True:  ret, frame = video\_capture.read()  if not ret:  break  frame = cv2.resize(frame, (1028, 500))  mask = bg\_subtractor.apply(frame)  \_, mask = cv2.threshold(mask, 245, 255, cv2.THRESH\_BINARY)  contours, \_ = cv2.findContours(mask, cv2.RETR\_EXTERNAL, cv2.CHAIN\_APPROX\_SIMPLE)  list = \[\]  for cnt in contours:  area = cv2.contourArea(cnt)  if area > 1500:  # cv2.drawContours(frame, \[cnt\], -1, (0, 255, 0), 2)  x, y, w, h = cv2.boundingRect(cnt)  list.append(\[x, y, w, h\])  bbox\_idx = tracker.update(list)  for bbox in bbox\_idx:  x1, y1, x2, y2, id = bbox  cx = int(x1 + x1 + x2) // 2  cy = int(y1 + y1 + y2) // 2  result = cv2.pointPolygonTest(np.array(area1, np.int32), ((cx, cy)), False)  if result >= 0:  er\[id\] = (cx, cy)  if id in er:  result1 = cv2.pointPolygonTest(np.array(area2, np.int32), ((cx, cy)), False)  if result1 >= 0:  cv2.rectangle(frame, (x1, y1), (x2 + x1, y2 + y1), (0, 255, 0), 3)  cvzone.putTextRect(frame, f'{id}', (cx, cy), 2, 2)  cv2.circle(frame, (cx, cy), 5, (0, 255, 0), \-1)  if counter1.count(id) == 0:  counter1.append(id)  result2 = cv2.pointPolygonTest(np.array(area2, np.int32), ((cx, cy)), False)  if result2 >= 0:  ex\[id\] = (cx, cy)  if id in ex:  result3 = cv2.pointPolygonTest(np.array(area1, np.int32), ((cx, cy)), False)  if result3 >= 0:  cv2.rectangle(frame, (x1, y1), (x2 + x1, y2 + y1), (0, 0, 255), 3)  cvzone.putTextRect(frame, f'{id}', (cx, cy), 2, 2)  cv2.circle(frame, (cx, cy), 5, (0, 255, 0), \-1)  if counter2.count(id) == 0:  counter2.append(id)  cv2.polylines(frame, \[np.array(area1, np.int32)\], True, (0, 0, 255), 2)  cv2.polylines(frame, \[np.array(area2, np.int32)\], True, (0, 0, 255), 2)  Enter = len(counter1)  Exit = len(counter2)  cvzone.putTextRect(frame, f'ENTER:-{Enter}', (50, 60), 2, 2)  cvzone.putTextRect(frame, f'EXIT:-{Exit}', (50, 130), 2, 2)  cv2.imshow('RGB', frame)  time.sleep(0.01)  if cv2.waitKey(1) & 0xFF == 27:  # Press 'Esc' to exit  break  # 释放资源, 关闭窗口  
video\_capture.release()  
cv2.destroyAllWindows()

结果如下。

/ 02 /

鸡蛋计数器

使用OpenCV和YOLOv8检测鸡蛋个数。

能够高效、准确、安全可靠的完成鸡蛋个数的统计工作。

基于yolov8-seg实例分割的实时检测。

具体代码如下。

import cv2  
from yolo\_segmentation import YOLOSEG  
import cvzone  
from tracker import \*  
import numpy as np  ys = YOLOSEG("best.pt")  my\_file = open("coco1.txt", "r")  
data = my\_file.read()  
class\_list = data.split("\\n")  cap = cv2.VideoCapture('egg.mp4')  
count = 0  def RGB(event, x, y, flags, param):  if event == cv2.EVENT\_MOUSEMOVE:  point = \[x, y\]  print(point)  cv2.namedWindow('RGB')  
cv2.setMouseCallback('RGB', RGB)  
tracker = Tracker()  
area = \[(434, 39), (453, 629), (473, 634), (456, 36)\]  
counter1 = \[\]  
while True:  ret, frame = cap.read()  if not ret:  break  frame = cv2.resize(frame, (1020, 700))  overlay = frame.copy()  alpha = 0.5  bboxes, classes, segmentations, scores = ys.detect(frame)  bbox\_idx = tracker.update(bboxes)  for bbox, seg in zip(bbox\_idx, segmentations):  x3, y3, x4, y4, id = bbox  cx = int(x3 + x4) // 2  cy = int(y3 + y4) // 2  result = cv2.pointPolygonTest(np.array(area, np.int32), ((cx, cy)), False)  if result >= 0:  #       cv2.rectangle(frame, (x3, y3), (x4, y4), (255, 0, 0), 2)  #           cv2.polylines(frame, \[seg\], True, (0, 0, 255), 4)  cv2.circle(frame, (cx, cy), 4, (0, 255, 0), \-1)  cv2.fillPoly(overlay, \[seg\], (0, 0, 255))  cv2.addWeighted(overlay, alpha, frame, 1 - alpha, 2, frame)  cvzone.putTextRect(frame, f'{id}', (x3, y3), 1, 1)  if counter1.count(id) == 0:  counter1.append(id)  cv2.polylines(frame, \[np.array(area, np.int32)\], True, (255, 0, 0), 2)  ca1 = len(counter1)  cvzone.putTextRect(frame, f'Egg: {ca1}', (50, 60), 2, 2)  cv2.imshow("RGB", frame)  if cv2.waitKey(1) & 0xFF == 27:  break  
cap.release()  
cv2.destroyAllWindows()

运行代码，结果如下。

/ 03 /

停车位计数器

使用OpenCV和YOLOv8检测停车场剩余车位。

提醒车主停车场各个区域的剩余车位信息。

使停车场车位管理更加规范有序，提高车位使用率。

import cv2  
import pandas as pd  
import numpy as np  
from ultralytics import YOLO  
import time  model = YOLO('yolov8s.pt')  def RGB(event, x, y, flags, param):  if event == cv2.EVENT\_MOUSEMOVE:  colorsBGR = \[x, y\]  print(colorsBGR)  cv2.namedWindow('RGB')  
cv2.setMouseCallback('RGB', RGB)  cap = cv2.VideoCapture('parking1.mp4')  my\_file = open("coco.txt", "r")  
data = my\_file.read()  
class\_list = data.split("\\n")  area1 = \[(52, 364), (30, 417), (73, 412), (88, 369)\]  area2 = \[(105, 353), (86, 428), (137, 427), (146, 358)\]  area3 = \[(159, 354), (150, 427), (204, 425), (203, 353)\]  area4 = \[(217, 352), (219, 422), (273, 418), (261, 347)\]  area5 = \[(274, 345), (286, 417), (338, 415), (321, 345)\]  area6 = \[(336, 343), (357, 410), (409, 408), (382, 340)\]  area7 = \[(396, 338), (426, 404), (479, 399), (439, 334)\]  area8 = \[(458, 333), (494, 397), (543, 390), (495, 330)\]  area9 = \[(511, 327), (557, 388), (603, 383), (549, 324)\]  area10 = \[(564, 323), (615, 381), (654, 372), (596, 315)\]  area11 = \[(616, 316), (666, 369), (703, 363), (642, 312)\]  area12 = \[(674, 311), (730, 360), (764, 355), (707, 308)\]  while True:  ret, frame = cap.read()  if not ret:  break  time.sleep(1)  frame = cv2.resize(frame, (1020, 500))  results = model.predict(frame)  #   print(results)  a = results\[0\].boxes.boxes  px = pd.DataFrame(a).astype("float")  #    print(px)  list1 = \[\]  list2 = \[\]  list3 = \[\]  list4 = \[\]  list5 = \[\]  list6 = \[\]  list7 = \[\]  list8 = \[\]  list9 = \[\]  list10 = \[\]  list11 = \[\]  list12 = \[\]  for index, row in px.iterrows():  #        print(row)  x1 = int(row\[0\])  y1 = int(row\[1\])  x2 = int(row\[2\])  y2 = int(row\[3\])  d = int(row\[5\])  c = class\_list\[d\]  if 'car' in c:  cx = int(x1 + x2) // 2  cy = int(y1 + y2) // 2  results1 = cv2.pointPolygonTest(np.array(area1, np.int32), ((cx, cy)), False)  if results1 >= 0:  cv2.rectangle(frame, (x1, y1), (x2, y2), (0, 255, 0), 2)  cv2.circle(frame, (cx, cy), 3, (0, 0, 255), \-1)  list1.append(c)  cv2.putText(frame, str(c), (x1, y1), cv2.FONT\_HERSHEY\_COMPLEX, 0.5, (255, 255, 255), 1)  results2 = cv2.pointPolygonTest(np.array(area2, np.int32), ((cx, cy)), False)  if results2 >= 0:  cv2.rectangle(frame, (x1, y1), (x2, y2), (0, 255, 0), 2)  cv2.circle(frame, (cx, cy), 3, (0, 0, 255), \-1)  list2.append(c)  results3 = cv2.pointPolygonTest(np.array(area3, np.int32), ((cx, cy)), False)  if results3 >= 0:  cv2.rectangle(frame, (x1, y1), (x2, y2), (0, 255, 0), 2)  cv2.circle(frame, (cx, cy), 3, (0, 0, 255), \-1)  list3.append(c)  results4 = cv2.pointPolygonTest(np.array(area4, np.int32), ((cx, cy)), False)  if results4 >= 0:  cv2.rectangle(frame, (x1, y1), (x2, y2), (0, 255, 0), 2)  cv2.circle(frame, (cx, cy), 3, (0, 0, 255), \-1)  list4.append(c)  results5 = cv2.pointPolygonTest(np.array(area5, np.int32), ((cx, cy)), False)  if results5 >= 0:  cv2.rectangle(frame, (x1, y1), (x2, y2), (0, 255, 0), 2)  cv2.circle(frame, (cx, cy), 3, (0, 0, 255), \-1)  list5.append(c)  results6 = cv2.pointPolygonTest(np.array(area6, np.int32), ((cx, cy)), False)  if results6 >= 0:  cv2.rectangle(frame, (x1, y1), (x2, y2), (0, 255, 0), 2)  cv2.circle(frame, (cx, cy), 3, (0, 0, 255), \-1)  list6.append(c)  results7 = cv2.pointPolygonTest(np.array(area7, np.int32), ((cx, cy)), False)  if results7 >= 0:  cv2.rectangle(frame, (x1, y1), (x2, y2), (0, 255, 0), 2)  cv2.circle(frame, (cx, cy), 3, (0, 0, 255), \-1)  list7.append(c)  results8 = cv2.pointPolygonTest(np.array(area8, np.int32), ((cx, cy)), False)  if results8 >= 0:  cv2.rectangle(frame, (x1, y1), (x2, y2), (0, 255, 0), 2)  cv2.circle(frame, (cx, cy), 3, (0, 0, 255), \-1)  list8.append(c)  results9 = cv2.pointPolygonTest(np.array(area9, np.int32), ((cx, cy)), False)  if results9 >= 0:  cv2.rectangle(frame, (x1, y1), (x2, y2), (0, 255, 0), 2)  cv2.circle(frame, (cx, cy), 3, (0, 0, 255), \-1)  list9.append(c)  results10 = cv2.pointPolygonTest(np.array(area10, np.int32), ((cx, cy)), False)  if results10 >= 0:  cv2.rectangle(frame, (x1, y1), (x2, y2), (0, 255, 0), 2)  cv2.circle(frame, (cx, cy), 3, (0, 0, 255), \-1)  list10.append(c)  results11 = cv2.pointPolygonTest(np.array(area11, np.int32), ((cx, cy)), False)  if results11 >= 0:  cv2.rectangle(frame, (x1, y1), (x2, y2), (0, 255, 0), 2)  cv2.circle(frame, (cx, cy), 3, (0, 0, 255), \-1)  list11.append(c)  results12 = cv2.pointPolygonTest(np.array(area12, np.int32), ((cx, cy)), False)  if results12 >= 0:  cv2.rectangle(frame, (x1, y1), (x2, y2), (0, 255, 0), 2)  cv2.circle(frame, (cx, cy), 3, (0, 0, 255), \-1)  list12.append(c)  a1 = (len(list1))  a2 = (len(list2))  a3 = (len(list3))  a4 = (len(list4))  a5 = (len(list5))  a6 = (len(list6))  a7 = (len(list7))  a8 = (len(list8))  a9 = (len(list9))  a10 = (len(list10))  a11 = (len(list11))  a12 = (len(list12))  o = (a1 + a2 + a3 + a4 + a5 + a6 + a7 + a8 + a9 + a10 + a11 + a12)  space = (12 - o)  print(space)  if a1 == 1:  cv2.polylines(frame, \[np.array(area1, np.int32)\], True, (0, 0, 255), 2)  cv2.putText(frame, str('1'), (50, 441), cv2.FONT\_HERSHEY\_COMPLEX, 0.5, (0, 0, 255), 1)  else:  cv2.polylines(frame, \[np.array(area1, np.int32)\], True, (0, 255, 0), 2)  cv2.putText(frame, str('1'), (50, 441), cv2.FONT\_HERSHEY\_COMPLEX, 0.5, (255, 255, 255), 1)  if a2 == 1:  cv2.polylines(frame, \[np.array(area2, np.int32)\], True, (0, 0, 255), 2)  cv2.putText(frame, str('2'), (106, 440), cv2.FONT\_HERSHEY\_COMPLEX, 0.5, (0, 0, 255), 1)  else:  cv2.polylines(frame, \[np.array(area2, np.int32)\], True, (0, 255, 0), 2)  cv2.putText(frame, str('2'), (106, 440), cv2.FONT\_HERSHEY\_COMPLEX, 0.5, (255, 255, 255), 1)  if a3 == 1:  cv2.polylines(frame, \[np.array(area3, np.int32)\], True, (0, 0, 255), 2)  cv2.putText(frame, str('3'), (175, 436), cv2.FONT\_HERSHEY\_COMPLEX, 0.5, (0, 0, 255), 1)  else:  cv2.polylines(frame, \[np.array(area3, np.int32)\], True, (0, 255, 0), 2)  cv2.putText(frame, str('3'), (175, 436), cv2.FONT\_HERSHEY\_COMPLEX, 0.5, (255, 255, 255), 1)  if a4 == 1:  cv2.polylines(frame, \[np.array(area4, np.int32)\], True, (0, 0, 255), 2)  cv2.putText(frame, str('4'), (250, 436), cv2.FONT\_HERSHEY\_COMPLEX, 0.5, (0, 0, 255), 1)  else:  cv2.polylines(frame, \[np.array(area4, np.int32)\], True, (0, 255, 0), 2)  cv2.putText(frame, str('4'), (250, 436), cv2.FONT\_HERSHEY\_COMPLEX, 0.5, (255, 255, 255), 1)  if a5 == 1:  cv2.polylines(frame, \[np.array(area5, np.int32)\], True, (0, 0, 255), 2)  cv2.putText(frame, str('5'), (315, 429), cv2.FONT\_HERSHEY\_COMPLEX, 0.5, (0, 0, 255), 1)  else:  cv2.polylines(frame, \[np.array(area5, np.int32)\], True, (0, 255, 0), 2)  cv2.putText(frame, str('5'), (315, 429), cv2.FONT\_HERSHEY\_COMPLEX, 0.5, (255, 255, 255), 1)  if a6 == 1:  cv2.polylines(frame, \[np.array(area6, np.int32)\], True, (0, 0, 255), 2)  cv2.putText(frame, str('6'), (386, 421), cv2.FONT\_HERSHEY\_COMPLEX, 0.5, (0, 0, 255), 1)  else:  cv2.polylines(frame, \[np.array(area6, np.int32)\], True, (0, 255, 0), 2)  cv2.putText(frame, str('6'), (386, 421), cv2.FONT\_HERSHEY\_COMPLEX, 0.5, (255, 255, 255), 1)  if a7 == 1:  cv2.polylines(frame, \[np.array(area7, np.int32)\], True, (0, 0, 255), 2)  cv2.putText(frame, str('7'), (456, 414), cv2.FONT\_HERSHEY\_COMPLEX, 0.5, (0, 0, 255), 1)  else:  cv2.polylines(frame, \[np.array(area7, np.int32)\], True, (0, 255, 0), 2)  cv2.putText(frame, str('7'), (456, 414), cv2.FONT\_HERSHEY\_COMPLEX, 0.5, (255, 255, 255), 1)  if a8 == 1:  cv2.polylines(frame, \[np.array(area8, np.int32)\], True, (0, 0, 255), 2)  cv2.putText(frame, str('8'), (527, 406), cv2.FONT\_HERSHEY\_COMPLEX, 0.5, (0, 0, 255), 1)  else:  cv2.polylines(frame, \[np.array(area8, np.int32)\], True, (0, 255, 0), 2)  cv2.putText(frame, str('8'), (527, 406), cv2.FONT\_HERSHEY\_COMPLEX, 0.5, (255, 255, 255), 1)  if a9 == 1:  cv2.polylines(frame, \[np.array(area9, np.int32)\], True, (0, 0, 255), 2)  cv2.putText(frame, str('9'), (591, 398), cv2.FONT\_HERSHEY\_COMPLEX, 0.5, (0, 0, 255), 1)  else:  cv2.polylines(frame, \[np.array(area9, np.int32)\], True, (0, 255, 0), 2)  cv2.putText(frame, str('9'), (591, 398), cv2.FONT\_HERSHEY\_COMPLEX, 0.5, (255, 255, 255), 1)  if a10 == 1:  cv2.polylines(frame, \[np.array(area10, np.int32)\], True, (0, 0, 255), 2)  cv2.putText(frame, str('10'), (649, 384), cv2.FONT\_HERSHEY\_COMPLEX, 0.5, (0, 0, 255), 1)  else:  cv2.polylines(frame, \[np.array(area10, np.int32)\], True, (0, 255, 0), 2)  cv2.putText(frame, str('10'), (649, 384), cv2.FONT\_HERSHEY\_COMPLEX, 0.5, (255, 255, 255), 1)  if a11 == 1:  cv2.polylines(frame, \[np.array(area11, np.int32)\], True, (0, 0, 255), 2)  cv2.putText(frame, str('11'), (697, 377), cv2.FONT\_HERSHEY\_COMPLEX, 0.5, (0, 0, 255), 1)  else:  cv2.polylines(frame, \[np.array(area11, np.int32)\], True, (0, 255, 0), 2)  cv2.putText(frame, str('11'), (697, 377), cv2.FONT\_HERSHEY\_COMPLEX, 0.5, (255, 255, 255), 1)  if a12 == 1:  cv2.polylines(frame, \[np.array(area12, np.int32)\], True, (0, 0, 255), 2)  cv2.putText(frame, str('12'), (752, 371), cv2.FONT\_HERSHEY\_COMPLEX, 0.5, (0, 0, 255), 1)  else:  cv2.polylines(frame, \[np.array(area12, np.int32)\], True, (0, 255, 0), 2)  cv2.putText(frame, str('12'), (752, 371), cv2.FONT\_HERSHEY\_COMPLEX, 0.5, (255, 255, 255), 1)  cv2.putText(frame, 'parking lots: ' + str(space), (23, 40), cv2.FONT\_HERSHEY\_PLAIN, 2, (255, 0, 255), 2)  cv2.imshow("RGB", frame)  if cv2.waitKey(1) & 0xFF == 27:  break  
cap.release()  
cv2.destroyAllWindows()  
# stream.stop()

运行代码，结果如下。

发现效果还不错~

/ 04 /

总结

以上操作，就是三个使用YOLOv8实现的计数视觉项目。

当然我们还可以通过预训练模型实现其它功能。

如果预训练模型的检测效果在你要使用的场景不太好，那就是需要加加数据了~