初学者如何在笔记本电脑上快速运行 YOLOv8 进行实时识别

随着人工智能技术的飞速进步，计算机视觉领域取得了长足的进步，物体检测、图像分割、姿态估计等任务得到了广泛的关注。YOLO（You Only Look Once）系列算法作为物体检测领域的杰出代表，以其速度快、准确率高而广受欢迎。本文将详细介绍如何在计算机上快速运行 YOLOv8 实现检测、分割和姿态估计。通过阅读本文，您将快速了解如何在 Windows 系统上安装、配置和运行 YOLOv8 进行实时识别，为实际应用打下坚实的基础。

安装环境

• 个人电脑：ThinkPad AMD
• 操作系统：Windows 11 x64
• 软件：Python 3.11
• 环境：Anaconda

1.下载Anaconda3：

网址：https://www.anaconda.com/download/success

2.安装Anaconda3

我选择的安装目录是：D:\Anaconda\Anaconda3。勾选‘将 Anaconda3 添加到我的 PATH 环境变量中’，将 Anaconda 添加到环境变量中。

3. 验证安装

请打开PowerShell并输入命令conda -V。

如果您没有看到错误：

‘conda’ 不是内部或外部命令，也不是可运行程序或批处理文件

表示安装成功。如果安装时没有将 Anaconda 添加到环境变量中，则需要手动添加 Anaconda 环境变量。

4.创建并激活虚拟环境

YOLOv8 可以使用官方提供的库，YOLOv10 的部署步骤与上述1-3步相同，详细部署说明请参考：https://github.com/THU-MIG/yolov10。

Bash
#create new environment
conda create -n yolo8test python==3.8
#activate environment
conda activate yolo8test 
# Install the ultralytics package from PyPI
pip install ultralytics
复制代码

不同模型性能比较

官网提供了多种尺寸的YOLOv8模型，参数对比如下表：

参数解释

以YOLOv8分割对象分割模型为例，这些模型的复杂度、参数数量、计算要求、准确率、速度等各有不同，适用于不同的应用场景。

物体检测

代码：

 Python
import cv2
import time
import torch
from ultralytics import YOLO
 
# load YOLOv8n
model = YOLO('yolov8n.pt')
cap = cv2.VideoCapture(0)
 
if not cap.isOpened():
    print("error")
    exit()
 
width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
 
prev_time = 0
fps = 0
while True:
 
    ret, frame = cap.read()
    
    if not ret:
        print("error")
        break
 
    results = model(frame, device='cpu')
 
    for result in results:
        boxes = result.boxes.xyxy.cpu().numpy()
        confidences = result.boxes.conf.cpu().numpy()
        class_ids = result.boxes.cls.cpu().numpy().astype(int)
        for i in range(len(boxes)):
            box = boxes[i]
            x1, y1, x2, y2 = map(int, box[:4])
            confidence = confidences[i]
            class_id = class_ids[i]
            label = result.names[class_id]
            cv2.rectangle(frame, (x1, y1), (x2, y2), (0, 255, 0), 2)
            cv2.putText(frame, f'{label} {confidence:.2f}', (x1, y1 - 10), 
                        cv2.FONT_HERSHEY_SIMPLEX, 0.9, (36, 255, 12), 2)
 
    curr_time = time.time()
    fps = 1 / (curr_time - prev_time)
    prev_time = curr_time
 
    cv2.putText(frame, f'FPS: {fps:.2f}', (10, 30), cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 255, 0), 2)
    cv2.imshow('YOLOv8n Real-time', frame)
    if cv2.waitKey(1) & 0xFF == ord('q'):
        break
 
cap.release()
cv2.destroyAllWindows()
复制代码

Bash
python yolov8ndet.py
复制代码

Python
import cv2
import time
# import torch
from ultralytics import YOLO
from PIL import Image
import numpy as np
 
model = YOLO('yolov8n-seg.pt')
cap = cv2.VideoCapture(0)
 
if not cap.isOpened():
    print("error")
    exit()
 
width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
 
 
prev_time = 0
fps = 0
 
while True:
 
    ret, frame = cap.read()
    if not ret:
        print("error")
        break
    results = model(frame, device='cpu')
    for result in results:
 
        pil_image = Image.fromarray(results[0].plot()[:, :, ::-1])
        frame = np.array(pil_image)
 
        boxes = result.boxes.xyxy.cpu().numpy()
        confidences = result.boxes.conf.cpu().numpy()
        class_ids = result.boxes.cls.cpu().numpy().astype(int)
 
        for i in range(len(boxes)):
            box = boxes[i]
            x1, y1, x2, y2 = map(int, box[:4])
            confidence = confidences[i]
            class_id = class_ids[i]
            label = result.names[class_id]
 
    curr_time = time.time()
    fps = 1 / (curr_time - prev_time)
    prev_time = curr_time
 
    cv2.putText(frame, f'FPS: {fps:.2f}', (10, 30), cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 255, 0), 2)
    cv2.imshow('YOLOv8n Real-time', frame)
   
 
    if cv2.waitKey(1) & 0xFF == ord('q'):
        break
 
cap.release()
cv2.destroyAllWindows()
复制代码

Bash
python yolov8nseg.py
复制代码

Python
import cv2
import time
from ultralytics import YOLO
from PIL import Image
import numpy as np
 
model = YOLO('yolov8n-pose.pt')
cap = cv2.VideoCapture(0)
 
width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
 
prev_time = 0
fps = 0
while True:
 
    ret, frame = cap.read()
    if not ret:
        print("error")
        break
 
 
    results = model(frame, device='cpu')
 
    for result in results:
        pil_image = Image.fromarray(results[0].plot())
        frame = np.array(pil_image)
        boxes = result.boxes.xyxy.cpu().numpy()
        confidences = result.boxes.conf.cpu().numpy()
        class_ids = result.boxes.cls.cpu().numpy().astype(int)
 
        for i in range(len(boxes)):
            box = boxes[i]
            x1, y1, x2, y2 = map(int, box[:4])
            confidence = confidences[i]
            class_id = class_ids[i]
            label = result.names[class_id]
    curr_time = time.time()
    fps = 1 / (curr_time - prev_time)
    prev_time = curr_time
    frame = cv2.resize(frame, (640, 480))  
 
    cv2.putText(frame, f'FPS: {fps:.2f}', (10, 30), cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 255, 0), 2)
    cv2.imshow('YOLOv8n Real-time', frame)
    if cv2.waitKey(1) & 0xFF == ord('q'):
        break
 
cap.release()
cv2.destroyAllWindows()
复制代码

 Bash
python yolov8npos.py
复制代码