【重要说明】
该系统以opencvsharp作图像处理,onnxruntime做推理引擎，使用CPU进行推理，适合有显卡或者没有显卡windows x64系统均可，不支持macOS和Linux系统，不支持x86的windows操作系统。由于采用CPU推理，要比GPU慢。为了适合大部分操作系统我们暂时只写了CPU推理源码，GPU推理源码后期根据需要可能会调整，目前只考虑CPU推理，主要是为了照顾现在大部分使用该源码是学生，很多人并没有显卡的电脑情况。
【算法介绍】
基于YOLOv8的课堂行为检测系统是现代教育技术的创新应用，该系统利用YOLOv8这一先进的深度学习算法，实现了对学生课堂行为的自动、高效和精准监测。YOLOv8在目标检测领域以其卓越的性能和速度著称，通过对学生上课视频或实时摄像头的输入进行深度分析，系统能够准确识别学生的多种行为，如举手、阅读、写作、使用手机、低头等。
该系统不仅提高了课堂监测的效率和准确性，还具备实时反馈功能，帮助教师即时调整教学策略，提升教学质量。通过图形化界面，教师可以方便地查看学生的行为数据，包括行为类型、发生时间、持续时间等，进而生成详细的行为分析报告。
此外，基于YOLOv8的课堂行为检测系统还具备多场景适用性，无论是传统教室还是在线教学环境，都能灵活应用。同时，系统收集的数据也可用于教育决策和个性化教学，为教育心理学、教学方法等方面的研究提供科学依据。
总之，基于YOLOv8的课堂行为检测系统以其高效、准确和实时的特点，为现代教育带来了革命性的变革，极大地提升了教学质量和学习效果。
【效果展示】
【测试环境】
windows10 x64系统
VS2019
netframework4.7.2
opencvsharp4.8.0
onnxruntime1.16.3
【模型可以检测出类别】
["Using_phone","bend","book","bow_head","hand-raising","phone","raise_head","reading","sleep","turn_head","upright","writing"]
【训练数据集】
https://download.csdn.net/download/FL1623863129/89699042
【部分实现源码】
using OpenCvSharp;
using System;
using System.Collections.Generic;
using System.ComponentModel;
using System.Data;
using System.Diagnostics;
using System.Drawing;
using System.IO;
using System.Linq;
using System.Text;
using System.Threading;
using System.Threading.Tasks;
using System.Windows.Forms;namespace FIRC
{public partial class Form1 : Form{public bool videoStart = false;//视频停止标志string weightsPath = Application.StartupPath + "\\weights";//模型目录string labelTxt= Application.StartupPath + "\\weights\\class_names.txt";//类别文件Yolov8Manager detetor = new Yolov8Manager();//推理引擎public Form1(){InitializeComponent();CheckForIllegalCrossThreadCalls = false;//线程更新控件不报错}private void LoadWeightsFromDir(){var di = new DirectoryInfo(weightsPath);foreach(var fi in di.GetFiles("*.onnx")){comboBox1.Items.Add(fi.Name);}if(comboBox1.Items.Count>0){comboBox1.SelectedIndex = 0;}else{tssl_show.Text = "未找到模型,请关闭程序，放入模型到weights文件夹!";tsb_pic.Enabled = false;tsb_video.Enabled = false;tsb_camera.Enabled = false;}}private void Form1_Load(object sender, EventArgs e){LoadWeightsFromDir();//从目录加载模型}public string GetResultString(Result result){Dictionary<string, int> resultDict = new Dictionary<string, int>();for (int i = 0; i < result.length; i++){if(resultDict.ContainsKey( result.classes[i]) ){resultDict[result.classes[i]]++;}else{resultDict[result.classes[i]]=1;}}var resultStr = "";foreach(var item in resultDict){resultStr += string.Format("{0}:{1}\n",item.Key,item.Value);}return resultStr;}private void tsb_pic_Click(object sender, EventArgs e){OpenFileDialog ofd = new OpenFileDialog();ofd.Filter = "*.*|*.bmp;*.jpg;*.jpeg;*.tiff;*.tiff;*.png";if (ofd.ShowDialog() != DialogResult.OK) return;tssl_show.Text = "正在检测中...";Task.Run(() => {var sw = new Stopwatch();sw.Start();Mat image = Cv2.ImRead(ofd.FileName);detetor.Confidence =Convert.ToSingle(numericUpDown1.Value);detetor.IOU = Convert.ToSingle(numericUpDown2.Value);var results=detetor.Inference(image);var resultImage = detetor.DrawImage(OpenCvSharp.Extensions.BitmapConverter.ToBitmap(image), results);sw.Stop();pb_show.Image = resultImage;tb_res.Text = GetResultString(results);tssl_show.Text = "检测已完成!总计耗时"+sw.Elapsed.TotalSeconds+"秒";});}public void VideoProcess(string videoPath){Task.Run(() => {detetor.Confidence = Convert.ToSingle(numericUpDown1.Value);detetor.IOU = Convert.ToSingle(numericUpDown2.Value);VideoCapture capture = new VideoCapture(videoPath);if (!capture.IsOpened()){tssl_show.Text="视频打开失败!";return;}Mat frame = new Mat();var sw = new Stopwatch();int fps = 0;while (videoStart){capture.Read(frame);if (frame.Empty()){Console.WriteLine("data is empty!");break;}sw.Start();var results = detetor.Inference(frame);var resultImg = detetor.DrawImage(frame,results);sw.Stop();fps = Convert.ToInt32(1 / sw.Elapsed.TotalSeconds);sw.Reset();Cv2.PutText(resultImg, "FPS=" + fps, new OpenCvSharp.Point(30, 30), HersheyFonts.HersheyComplex, 1.0, new Scalar(255, 0, 0), 3);//显示结果pb_show.Image = OpenCvSharp.Extensions.BitmapConverter.ToBitmap(resultImg);tb_res.Text = GetResultString(results);Thread.Sleep(5);}capture.Release();pb_show.Image = null;tssl_show.Text = "视频已停止!";tsb_video.Text = "选择视频";});}public void CameraProcess(int cameraIndex=0){Task.Run(() => {detetor.Confidence = Convert.ToSingle(numericUpDown1.Value);detetor.IOU = Convert.ToSingle(numericUpDown2.Value);VideoCapture capture = new VideoCapture(cameraIndex);if (!capture.IsOpened()){tssl_show.Text = "摄像头打开失败!";return;}Mat frame = new Mat();var sw = new Stopwatch();int fps = 0;while (videoStart){capture.Read(frame);if (frame.Empty()){Console.WriteLine("data is empty!");break;}sw.Start();var results = detetor.Inference(frame);var resultImg = detetor.DrawImage(frame, results);sw.Stop();fps = Convert.ToInt32(1 / sw.Elapsed.TotalSeconds);sw.Reset();Cv2.PutText(resultImg, "FPS=" + fps, new OpenCvSharp.Point(30, 30), HersheyFonts.HersheyComplex, 1.0, new Scalar(255, 0, 0), 3);//显示结果pb_show.Image = OpenCvSharp.Extensions.BitmapConverter.ToBitmap(resultImg);tb_res.Text = GetResultString(results);Thread.Sleep(5);}capture.Release();pb_show.Image = null;tssl_show.Text = "摄像头已停止!";tsb_camera.Text = "打开摄像头";});}private void tsb_video_Click(object sender, EventArgs e){if(tsb_video.Text=="选择视频"){OpenFileDialog ofd = new OpenFileDialog();ofd.Filter = "视频文件(*.*)|*.mp4;*.avi";if (ofd.ShowDialog() != DialogResult.OK) return;videoStart = true;VideoProcess(ofd.FileName);tsb_video.Text = "停止";tssl_show.Text = "视频正在检测中...";}else{videoStart = false;}}private void tsb_camera_Click(object sender, EventArgs e){if (tsb_camera.Text == "打开摄像头"){videoStart = true;CameraProcess(0);tsb_camera.Text = "停止";tssl_show.Text = "摄像头正在检测中...";}else{videoStart = false;}}private void tsb_exit_Click(object sender, EventArgs e){videoStart = false;this.Close();}private void trackBar1_Scroll(object sender, EventArgs e){numericUpDown1.Value = Convert.ToDecimal(trackBar1.Value / 100.0f);}private void trackBar2_Scroll(object sender, EventArgs e){numericUpDown2.Value = Convert.ToDecimal(trackBar2.Value / 100.0f);}private void numericUpDown1_ValueChanged(object sender, EventArgs e){trackBar1.Value = (int)(Convert.ToSingle(numericUpDown1.Value) * 100);}private void numericUpDown2_ValueChanged(object sender, EventArgs e){trackBar2.Value = (int)(Convert.ToSingle(numericUpDown2.Value) * 100);}private void comboBox1_SelectedIndexChanged(object sender, EventArgs e){tssl_show.Text="加载模型:"+comboBox1.Text;detetor.LoadWeights(weightsPath+"\\"+comboBox1.Text,labelTxt);tssl_show.Text = "模型加载已完成!";}}
}
【使用步骤】
使用步骤：
（1）首先根据官方框架https://github.com/ultralytics/ultralytics安装教程安装好yolov8环境，并根据官方export命令将自己pt模型转成onnx模型
（2）使用vs2019打开sln项目，选择x64 release并且修改一些必要的参数，比如输入shape等，点击运行即可查看最后效果
特别注意如果运行报错了，请参考我的博文进行重新引用我源码的DLL：[C#]opencvsharp报错System.Memory，Version=4.0.1.2，Culture=neutral，PublicKeyToken=cc7b13fcd2ddd51“版本高于所引_未能加载文件或程序集“system.memory, version=4.0.1.2, culture-CSDN博客
【提供文件】
C#源码
yolov8n.onnx模型（不提供pytorch模型）
训练的map,P,R曲线图(在weights\results.png)
测试图片（在test_img文件夹下面）
【源码下载地址】
关注下方名片并回复【firc10】即可获取下载方式
