实战 | 基于YOLOv9和OpenCV实现车辆跟踪计数(步骤 + 源码)

导 读

本文主要介绍使用YOLOv9和OpenCV实现车辆跟踪计数(步骤 + 源码)。

实现步骤

监控摄像头可以有效地用于各种场景下的车辆计数和交通流量统计。先进的计算机视觉技术(例如对象检测和跟踪)可应用于监控录像,以识别和跟踪车辆在摄像机视野中移动。

【1】安装ultralytics,因为它拥有直接使用 YoloV9 预训练模型的方法。

复制代码
pip install ultralytics

【2 】完成后,就可以创建跟踪器函数来跟踪对象了。我们只是为此创建了一个名为tracker.py的python文件。

复制代码
import math
class CustomTracker:    def __init__(self):        # Store the center positions of the objects        self.custom_center_points = {}        # Keep the count of the IDs        # each time a new object id detected, the count will increase by one        self.custom_id_count = 0
    def custom_update(self, custom_objects_rect):        # Objects boxes and ids        custom_objects_bbs_ids = []
        # Get center point of new object        for custom_rect in custom_objects_rect:            x, y, w, h = custom_rect            cx = (x + x + w) // 2            cy = (y + y + h) // 2
            # Find out if that object was detected already            same_object_detected = False            for custom_id, pt in self.custom_center_points.items():                dist = math.hypot(cx - pt[0], cy - pt[1])
                if dist < 35:                    self.custom_center_points[custom_id] = (cx, cy)                    custom_objects_bbs_ids.append([x, y, w, h, custom_id])                    same_object_detected = True                    break
            # New object is detected we assign the ID to that object            if same_object_detected is False:                self.custom_center_points[self.custom_id_count] = (cx, cy)                custom_objects_bbs_ids.append([x, y, w, h, self.custom_id_count])                self.custom_id_count += 1
        # Clean the dictionary by center points to remove IDS not used anymore        new_custom_center_points = {}        for custom_obj_bb_id in custom_objects_bbs_ids:            _, _, _, _, custom_object_id = custom_obj_bb_id            center = self.custom_center_points[custom_object_id]            new_custom_center_points[custom_object_id] = center
        # Update dictionary with IDs not used removed        self.custom_center_points = new_custom_center_points.copy()        return custom_objects_bbs_ids

**【3】编写车辆计数的主要代码。**​​​​​​​

复制代码
# Import the Librariesimport cv2import pandas as pdfrom ultralytics import YOLOfrom tracker import *

导入所有必要的库后,就可以导入模型了。我们不必从任何存储库下载模型。Ultralytics 做得非常出色,让我们可以更轻松地直接下载它们。

复制代码
model=YOLO('yolov9c.pt')

这会将 yolov9c.pt 模型下载到当前目录中。该模型已经在由 80 个不同类别组成的 COCO 数据集上进行了训练。现在让我们指定类:​​​​​​​

复制代码
class_list = ['person', 'bicycle', 'car', 'motorcycle', 'airplane', 'bus',              'train', 'truck', 'boat', 'traffic light', 'fire hydrant', 'stop sign', 'parking meter',              'bench', 'bird', 'cat', 'dog', 'horse', 'sheep', 'cow', 'elephant', 'bear', 'zebra', 'giraffe',              'backpack', 'umbrella', 'handbag', 'tie', 'suitcase', 'frisbee', 'skis', 'snowboard', 'sports ball', 'kite',              'baseball bat', 'baseball glove', 'skateboard', 'surfboard', 'tennis racket', 'bottle', 'wine glass', 'cup',              'fork', 'knife', 'spoon', 'bowl', 'banana', 'apple', 'sandwich', 'orange', 'broccoli', 'carrot', 'hot dog',              'pizza', 'donut', 'cake', 'chair', 'couch', 'potted plant', 'bed', 'dining table', 'toilet', 'tv', 'laptop',              'mouse', 'remote', 'keyboard', 'cell phone', 'microwave', 'oven', 'toaster', 'sink', 'refrigerator', 'book',              'clock', 'vase', 'scissors', 'teddy bear', 'hair drier', 'toothbrush']

现在,下一步是加载您要使用的视频。​​​​​​​

复制代码
tracker=CustomTracker()count=0
cap = cv2.VideoCapture('traffictrim.mp4')
# Get video propertiesfps = int(cap.get(cv2.CAP_PROP_FPS))width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
# Create VideoWriter object to save the modified framesoutput_video_path = 'output_video.mp4'fourcc = cv2.VideoWriter_fourcc(*'mp4v')  # You can use other codecs like 'XVID' based on your systemout = cv2.VideoWriter(output_video_path, fourcc, fps, (width, height))

在这里,我们在加载视频后获取视频属性,因为它们对于使用计数器重新创建视频并最终将其存储在本地非常有用。​​​​​​​

复制代码
# Looping over each frame and Performing the Detection
down = {}counter_down = set()while True:    ret, frame = cap.read()    if not ret:        break    count += 1
    results = model.predict(frame)
    a = results[0].boxes.data    a = a.detach().cpu().numpy()    px = pd.DataFrame(a).astype("float")    # print(px)
    list = []
    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:            list.append([x1, y1, x2, y2])
    bbox_id = tracker.custom_update(list)    # print(bbox_id)    for bbox in bbox_id:        x3, y3, x4, y4, id = bbox        cx = int(x3 + x4) // 2        cy = int(y3 + y4) // 2        # cv2.circle(frame,(cx,cy),4,(0,0,255),-1) #draw ceter points of bounding box        # cv2.rectangle(frame, (x3, y3), (x4, y4), (0, 255, 0), 2)  # Draw bounding box        # cv2.putText(frame,str(id),(cx,cy),cv2.FONT_HERSHEY_COMPLEX,0.8,(0,255,255),2)
        y = 308        offset = 7
        ''' condition for red line '''        if y < (cy + offset) and y > (cy - offset):            ''' this if condition is putting the id and the circle on the object when the center of the object touched the red line.'''
            down[id] = cy  # cy is current position. saving the ids of the cars which are touching the red line first.            # This will tell us the travelling direction of the car.            if id in down:                cv2.circle(frame, (cx, cy), 4, (0, 0, 255), -1)                #cv2.putText(frame, str(id), (cx, cy), cv2.FONT_HERSHEY_COMPLEX, 0.8, (0, 255, 255), 2)                counter_down.add(id)
                # # line    text_color = (255, 255, 255)  # white color for text    red_color = (0, 0, 255)  # (B, G, R)
    # print(down)    cv2.line(frame, (282, 308), (1004, 308), red_color, 3)  # starting cordinates and end of line cordinates    cv2.putText(frame, ('red line'), (280, 308), cv2.FONT_HERSHEY_SIMPLEX, 0.5, text_color, 1, cv2.LINE_AA)

    downwards = (len(counter_down))    cv2.putText(frame, ('Vehicle Counter - ') + str(downwards), (60, 40), cv2.FONT_HERSHEY_SIMPLEX, 0.5, red_color, 1,                cv2.LINE_AA)
    cv2.line(frame,(282,308),(1004,308),red_color,3)  #  starting cordinates and end of line cordinates    cv2.putText(frame,('red line'),(280,308),cv2.FONT_HERSHEY_SIMPLEX, 0.5, text_color, 1, cv2.LINE_AA)        # This will write the Output Video to the location specified above    out.write(frame)

在上面的代码中,我们循环遍历视频中的每个帧,然后进行检测。然后,由于我们仅对车辆进行计数,因此仅过滤掉汽车的检测结果。

之后,我们找到检测到的车辆的中心,然后在它们穿过人工创建的红线时对它们进行计数。我们可以在下面的视频快照中清楚地看到它们。

我们可以看到,当车辆越过红线时,视频左上角的计数器不断增加。

相关推荐
Y1nhl28 分钟前
搜广推校招面经八十一
开发语言·人工智能·pytorch·深度学习·机器学习·推荐算法·搜索算法
胡攀峰28 分钟前
第12章 微调生成模型
人工智能·大模型·llm·sft·强化学习·rlhf·指令微调
yuanlaile29 分钟前
AI大模型自然语言处理能力案例演示
人工智能·ai·自然语言处理
YiSLWLL30 分钟前
使用Tauri 2.3.1+Leptos 0.7.8开发桌面小程序汇总
python·rust·sqlite·matplotlib·visual studio code
小白白搭建33 分钟前
WordPress AI 原创文章自动生成插件 24小时全自动生成SEO原创文章 | 多语言支持 | 智能配图与排版
人工智能
Jamence35 分钟前
多模态大语言模型arxiv论文略读(三十九)
人工智能·语言模型·自然语言处理
ai大模型木子1 小时前
嵌入模型(Embedding Models)原理详解:从Word2Vec到BERT的技术演进
人工智能·自然语言处理·bert·embedding·word2vec·ai大模型·大模型资料
花酒锄作田1 小时前
[flask]自定义请求日志
python·flask
SsummerC3 小时前
【leetcode100】组合总和Ⅳ
数据结构·python·算法·leetcode·动态规划
Tandy12356_3 小时前
Godot开发2D冒险游戏——第一节:主角登场!
python·游戏引擎·godot