yolov8识别网球场

#!/usr/bin/env python # -*- coding:utf-8 -*- #This Python codes are created by Li hexi for undergraduate student course --maxhine learning #Any copy of the codes is illegal without the author' permission.2021-9-3 #--------This program uses to crawl and download images on some websites ------ import math import imghdr from distutils.command.clean import clean import cv2 import re import time,os import urllib import requests import numpy as np import tkinter as tk import tkinter.filedialog as file import threading as thread import tkinter.messagebox as mes import tkinter.simpledialog as simple from PIL import Image, ImageDraw, ImageFont from tkinter import scrolledtext as st import argostranslate.translate as Translate from h5py.h5a import delete from pypinyin import pinyin, Style from pypinyin import lazy_pinyin, Style from matplotlib import pyplot as plt from matplotlib.backends.backend_tkagg import FigureCanvasTkAgg#导入在tkinter 内嵌入的画布子窗口 from matplotlib.backend_bases import MouseEvent from tkinter import scrolledtext as st from tkinter import ttk from ultralytics import YOLO MainWin = tk.Tk() #常数列表 #FrameWidth=600 #FrameHeight=300 FileLength=0 TagFlag=False TargetVector=[] CrawlFlag=False #ObjectName = ['人','电脑', '手机', '鼠标', '键盘'] #ObjectName=['人','电脑', '键盘', '鼠标', '订书机', '手机', '书', '台灯','杯子','电水壶'] Name80 = ['人', '自行车', '轿车', '摩托', '飞机', '巴士', '火车', '卡车', '船', '交通灯', '消防栓', '停止标志', '咪表', '长凳', '鸟', '猫', '狗', '马', '羊', '奶牛', '象', '熊', '斑马', '长颈鹿', '背包', '伞', '手袋', '领带', '手提箱', '飞碟', '滑板', '滑雪板', '运动球', '风筝', '棒球杆', '棒球手套', '滑板', '冲浪板', '网球拍', '瓶子', '酒杯', '杯子', '叉子', '刀', '勺子', '碗', '香蕉', '苹果', '三文治', '桔子', '西兰花', '胡萝卜', '热狗', '披萨', '甜甜圈', '蛋糕', '椅子', '沙发', '盆景', '床', '餐桌', '厕所', '监视器', '电脑', '鼠标', '遥控器', '键盘', '手机', '微波炉', '烤箱', '烤面包机', '洗碗槽', '冰箱', '书', '钟', '花瓶', '剪刀', '泰迪熊', '干发器', '牙刷'] Name10=['人','电脑','鼠标', '键盘', '订书机', '手机', '书', '台灯','杯子','电水壶'] Name2=['人','电脑'] ObjectName=Name80.copy() MainWin.title('目标检测系统智能体平台') MainWin.geometry('1300x700') MainWin.resizable(width=False,height=False) #========函数区开始=========== def callback1(): filename = file.askopenfilename(title='打开文件名字', initialdir="\image", filetypes=[('jpg文件', '*.jpg'), ('png文件', '.png')]) picshow(filename) def callback2(): #import argostranslate.package #argostranslate.package.update_package_index() #available_packages = argostranslate.package.get_available_packages() #package_to_install = next(p for p in available_packages if p.from_code == "zh" and p.to_code == "en") #argostranslate.package.install_from_path(package_to_install.download()) #text = Translate.translate("蔬菜和水果", "zh", "en") #FileName=text.replace(" ","") text = '紫荆花' pinyin_str = ''.join(lazy_pinyin(text)) print(pinyin_str) # ni hao #print(FileName) def picshow(filename): global GI I1 = cv2.imread(filename) I2 = cv2.resize(I1, (WW, WH)) cv2.imwrite('ImageFile/temp.png', I2) I3 = tk.PhotoImage(file='ImageFile/temp.png') L1 = tk.Label(InputFrame, image=I3) L1.grid(row=1, column=0, columnspan=2, padx=40) GI = I2 MainWin.update() #******************************** #******************************** #以下下是文件操作代码区************* #******************************** #******************************** def LoadImage(): global TargetVector, GI path=os.getcwd() filename = file.askopenfilename(title='打开文件名字', initialdir=path, filetypes=[('jpg文件', '*.jpg'), ('png文件', '.png')]) if len(filename) == 0: return I1 = cv2.imread(filename) I2 = cv2.resize(I1, (WW, WH)) GI = I2 # print(np.shape(GI)) cv2.imwrite('image/temp.png', I2) I3 = tk.PhotoImage(file='image/temp.png') L1 = tk.Label(InputFrame, image=I3) L1.grid(row=1, column=0, columnspan=2, padx=40) s = np.shape(I1) MainWin.mainloop() return def SaveImage(): return #******************************** #******************************** #以下下是图像爬取代码区************* #******************************** #******************************** def StartCrawling(): global CrawlFlag if KeyStr.get()=="": return CrawlFlag=True Thd1 = thread.Thread(target=CrawlPicture, args=(300, 500,)) Thd1.setDaemon(True) Thd1.start() return PauseFlag=False def StopCrawling(): global CrawlFlag CrawlFlag = False return def Suspending(): return def Resuming(): return #******************************** #******************************** #以下下是图像清洗代码区************* #******************************** #******************************** std=None def CleanImage(): global img, recimg, sw, sh, FileLength, pathname, FileNum, filelist, std, ax if std != None: std.destroy() std = None fig.clear() draw_set.draw() OutputFrame.update() curpath = os.getcwd() pathname = file.askdirectory(title='图像文件目录', initialdir=curpath) if len(pathname) == 0: return filelist = os.listdir(pathname) FileLength = len(filelist) DeletFile=[] fig.clear() ax = fig.add_subplot(1, 1, 1, position=[0, 0, 1, 1]) for i in range(FileLength): fstr = pathname + '/' + filelist[i] if imghdr.what(fstr) is not None: #print(fstr) img = cv2.imread(fstr) if np.shape(img) == (): DeletFile.append(fstr) else: s = np.shape(img) sw = s[0] sh = s[1] if s[0] > 1024 or s[1] > 1024: sv = 1024.0 / max(s[0], s[1]) img = cv2.resize(img, (int(sv * s[1]), int(sv * s[0]))) cv2.imwrite(fstr, img) img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB) ax.imshow(img) draw_set.draw() OutputFrame.update() else: DeletFile.append(fstr) if len(DeletFile) !=0: for i in range(len(DeletFile)): fstr = DeletFile[i] os.remove(fstr) return def ScanImage(): global img, recimg, sw, sh, FileLength, pathname, FileNum, filelist, std, ax if std!=None: std.destroy() std=None fig.clear() draw_set.draw() OutputFrame.update() curpath = os.getcwd() pathname = file.askdirectory(title='图像文件目录', initialdir=curpath) if len(pathname) == 0: return filelist = os.listdir(pathname) FileLength = len(filelist) for i in range(FileLength): fstr = pathname + '/' + filelist[i] img = cv2.imread(fstr) if np.shape(img) == (): os.remove(fstr) filelist = os.listdir(pathname) FileLength = len(filelist) FileNum = 0 fstr = pathname + '/' + filelist[FileNum] img = cv2.imread(fstr) s = np.shape(img) sw = s[0] sh = s[1] if s[0] > 1024 or s[1] > 1024: sv = 1024.0 / max(s[0], s[1]) img = cv2.resize(img, (int(sv * s[1]), int(sv *s[0]))) recimg = np.zeros((sw, sh, 3), dtype=np.uint8) s = np.shape(img) sw = s[0] sh = s[1] img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB) fig.clear() ax = fig.add_subplot(1, 1, 1, position=[0, 0, 1, 1]) ax.imshow(img) draw_set.draw() recimg = np.zeros((WH, WW, 3), dtype=np.uint8) fig.canvas.mpl_connect('scroll_event', on_ax_scroll) fig.canvas.mpl_connect('button_press_event', on_ax_click) fig.canvas.mpl_connect('motion_notify_event', on_ax_motion) OutputFrame.update() return def RenameImage(): return def NPZfile(): return #******************************** #******************************** #以下下是图像标注代码区************** #******************************** #******************************** def LoadClassName(): return #批量区域标定 def LabelBatch(): global img, recimg, sw, sh, FileLength, pathname, FileNum, filelist, std2,ax2 global Mflag fig2.clear() draw_set2.draw() InputFrame.update() Lab3.config(text="图片标注结果输出") std2=st.ScrolledText(InputFrame,width=60,height=22,foreground="blue", font=("隶书",12)) std2.grid(row=1, column=0,columnspan=2,padx=10) std2.config(foreground="black", relief="solid",font=("宋体", 12)) std2.delete(0.0, tk.END) InputFrame.update() cwd = os.getcwd() pathname = file.askdirectory(title='图像文件目录', initialdir=cwd) if len(pathname) == 0: return filelist = os.listdir(pathname) FileLength = len(filelist) FileNum = 0 DeletFile=[] for i in range(FileLength): fstr = pathname + '/' + filelist[i] if imghdr.what(fstr) is not None: #print(fstr) img = cv2.imread(fstr) if np.shape(img) == (): DeletFile.append(fstr) else: fstr = DeletFile[i] DeletFile.append(fstr) for i in range(len(DeletFile)): os.remove(fstr) filelist = os.listdir(pathname) FileLength = len(filelist) FileNum = 0 fstr = pathname + '/' + filelist[FileNum] img = cv2.imread(fstr) s = np.shape(img) if s[0] > 1024 or s[1] > 1024: sv = 1024.0 / max(s[0], s[1]) img = cv2.resize(img, (int(sv * s[1]), int(sv * s[0]))) img = cv2.resize(img, (WW, WH)) recimg = np.zeros(img.shape, dtype=np.uint8) s = np.shape(img) sh = s[0] sw = s[1] img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB) fig.clear() ax = fig.add_subplot(1, 1, 1, position=[0, 0, 1, 1]) ax.imshow(img) draw_set.draw() recimg = np.zeros((WH, WW, 3), dtype=np.uint8) fig.canvas.mpl_connect('scroll_event', on_ax_scroll) fig.canvas.mpl_connect('button_press_event', on_ax_click) fig.canvas.mpl_connect('motion_notify_event', on_ax_motion) OutputFrame.update() but7.config(state=tk.DISABLED) but8.config(state=tk.DISABLED) Mflag=1 return def LabelEnd(): global LabelData, BoxData if len(BoxData) != 0: LabelData.append(BoxData.copy()) but8.config(state=tk.ACTIVE) return def LabelSave(): global ImageData, LabelData, BoxData cwd = os.getcwd() dir1 = cwd + "\image" if not os.path.exists(dir1): os.mkdir(dir1) startnum = len(os.listdir(dir1)) for i in range(len(ImageData)): fname = dir1 + "\myimage" + str(i + startnum) + ".jpg" myim = cv2.cvtColor(ImageData[i], cv2.COLOR_RGB2BGR) cv2.imwrite(fname, myim) dir1 = cwd + "\label" if not os.path.exists(dir1): os.mkdir(dir1) for i in range(len(ImageData)): fname = dir1 + "\myimage" + str(i + startnum) + ".txt" with open(fname, 'w', encoding='utf-8') as file: for j in range(len(LabelData[i])): txt = str(LabelData[i][j][0]) + " " + str(LabelData[i][j][1]) + " " + str( LabelData[i][j][2]) + " " + str(LabelData[i][j][3]) + " " + str(LabelData[i][j][4]) file.write(txt + '\n') # 添加换行符 file.close() ClearAnn() return #******************************** #******************************** #以下下是图像训练代码区************* #******************************** #******************************** def TrainSetting(): return def TrainStarting(): TrainYoloV8() return def TrainResultShow(): return def SaveTrainResult(): return #******************************** #******************************** #以下下是图像测试代码区************* #******************************** #******************************** def SelectImage(): LoadImage() return def SingleImageTest(): Yolov8SingleDetect() return def ObjectTracking(): Yolov8Detect() return def SaveTestResult(): return #******************************** #******************************** #以下下是操作帮助代码区************* #******************************** #******************************** def HelpPicCrawl(): global std if std != None: std.destroy() fig.clear() draw_set.draw() OutputFrame.update() std = st.ScrolledText(OutputFrame, width=60, height=22, foreground="blue", font=("隶书", 12)) std.grid(row=1, column=0, padx=10) std.config(relief="solid") #std.config(foreground="black", relief="solid", font=("宋体", 13)) std.delete(0.0, tk.END) Lab3.config(text="图片爬取操作指导") HelpStr1="图像爬取操作步骤：\r\n 1)在主题框输入要爬取的主题提示词" \ "\r\n 2)单击《开始爬取》按钮，开始爬取图片，并存在当前目录下" \ "\r\n 3)单击《停止爬取》按钮，停止爬取图片，但爬完当前页才结束"\ "\r\n 4)爬取的图像存在主题词拼音目录下" std.delete(0.0,tk.END) std.insert(tk.END,HelpStr1) return def HelpAnnSave(): return def HelpBatchAnn(): global std if std!=None: std.destroy() fig.clear() draw_set.draw() OutputFrame.update() std = st.ScrolledText(OutputFrame, width=60, height=22, foreground="blue", font=("隶书", 12)) std.config(relief="solid") std.grid(row=1, column=0, padx=10) std.delete(0.0, tk.END) Lab3.config(text="区域标注操作指导") HelpStr3 = "图像区域标注操作步骤：\r\n 1)单击《方框标注》按钮，弹出目录选择对话框，选择图片目录" \ "\r\n 2)计算机会对该目录的图像文件扫描，清洗掉一些格式不对的非图像文件，时间较长" \ "\r\n 3)显示目录下第1幅图像，通过鼠标滚轮，快速浏览目录下的所有图片" \ "\r\n 4)停在所选图片，在要标注的物体左上角按下鼠标左键，选择拉框起点" \ "\r\n 5)按下鼠标左键不放，拖动鼠标拉框，将要标注的物体画到矩形框内" \ "\r\n 6)点击鼠标右键，激活目标选项，滚动鼠标滚轮，浏览目标名称，并停在所选名称" \ "\r\n 7)点击鼠标左键，确认选项，并在文本框内显示类别ID和矩形几何参数" \ "\r\n 8)如果本张图片还有要标定的目标，则重复步骤4-7，继续标注" \ "\r\n 9)更换图片和浏览一样滚动鼠标滚轮，选择图片，然后重复步骤4-7" \ "\r\n 10)结束本次标定，单击《结束标注》按钮" \ "\r\n 11)保存标定结果，单击《保存标注》按钮"\ "\r\n 12)标定的图像保存在\image目录下，标签保存在\label目录下" std.delete(0.0, tk.END) #std.config(foreground="blue", font=("楷体", 12)) std.insert(tk.END, HelpStr3) ########################################## def TagPicture(): return def SpareBut(): global ImageData for i in range(len(ImageData)): cv2.imshow("ls",ImageData[i]) cv2.waitKey(200) return def GetPageURL(URLStr): #获取一个页面的所有图片的URL+下页的URL if not URLStr: print('现在是最后一页啦！爬取结束') return [], '' try: header = {"User-Agent": "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_7_2) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/27.0.1453.93 Safari/537.36"} html = requests.get(URLStr,headers=header)#,verify=False) html.encoding = 'utf-8' html = html.text except Exception as e: print("err=",str(e)) ImageURL = [] NextPageURL = '' return ImageURL, NextPageURL ImageURL = re.findall(r'"objURL":"(.*?)",', html, re.S) #print("ImageURL",ImageURL) NextPageURLS = re.findall(re.compile(r'下一页'), html, flags=0) if NextPageURLS: NextPageURL = 'http://image.baidu.com' + NextPageURLS[0] else: NextPageURL='' return ImageURL, NextPageURL ImageCount=0 def DownLoadImage(pic_urls): """给出图片链接列表, 下载所有图片""" global ImageCount,ImageFilePath, CrawlFlag #print(ImageFilePath) for i,pic_url in enumerate(pic_urls): if not CrawlFlag: return try: pic = requests.get(pic_url, timeout=6) ImageCount=ImageCount+1 #print(ImageCount) string = ImageFilePath+str(ImageCount)+".jpg" with open(string, 'wb') as f: f.write(pic.content) FileStr.set("已下载第"+str(ImageCount)+"图片") DownAddrStr.set(str(pic_url)) #print('成功下载第%s张图片: %s' % (str(i + 1), str(pic_url))) except Exception as e: FileStr.set("下载第"+(str(ImageCount)+"张图片失败")) ImageCount-=1 DownAddrStr.set(e) continue ImageFilePath='' def CrawlPicture(v1,v2): global CrawlFlag,PauseFlag,ImageFilePath,ImageCount while not CrawlFlag: time.sleep(1) ImageCount=0 str1 = os.getcwd() str2 = KeyStr.get() str3 = ''.join(lazy_pinyin(str2)) if str3 == '': str3 = "temp" ImageFilePath = str1 + "\\" + str3+"\\" #print(ImageFilePath) if not os.path.exists(ImageFilePath): os.makedirs(ImageFilePath) else: while os.path.exists(ImageFilePath): str3 = str3 + str(np.random.randint(1, 10000)) ImageFilePath = str1 + "\\" + str3 + "\\" os.makedirs(ImageFilePath) BaiduURL=r'https://image.baidu.com/search/flip?tn=baiduimage&ps=1&ct=201326592&lm=-1&cl=2&nc=1&ie=utf-8&word=' keyword=KeyStr.get() FirstURL=BaiduURL+urllib.parse.quote(keyword,safe='/') ImageURL, NextPageURL= GetPageURL(FirstURL) PageCount = 0 # 累计翻页数 while CrawlFlag: ImageURL, NextPageURL = GetPageURL(NextPageURL) PageCount += 1 CountStr.set(str(PageCount)) if ImageURL!=[]: DownLoadImage(list(set(ImageURL))) if NextPageURL== '': CrawlFlag=False def SetTarget(): global TagFlag, TargetVector TagFlag = True for i in range(10): if (TargetV.get() == i): TargetVector = np.zeros(10) TargetVector[i] = 1.0 print(TargetVector) #按“图像预处理”键，将弹出一个文件目录选择框，你选择一个目录，将列出此目录下的第一个文件，显示在“image"窗口 #然后就可以对此目录下的文件进行批处理 #将鼠标移到“image"窗口，利用鼠标橡皮筋功能裁剪图像，左键按下选择裁剪起点point1 #左键按下并拖动鼠标产生橡皮筋功能的矩形框，抬起选择结束 #点击右键将所选框的图像剪裁出来，并显示新的剪裁窗口”crop" #鼠标移到“crop"窗口，双击左键将将用这个剪裁的图像替代原图像 #鼠标移到“crop"窗口，双击右键将这个剪裁的图像取名另存盘，等于增加一幅图像 #连续选择图像功能，将鼠标移至”image",滚动鼠标的滚轮，在“image"窗口将连续滚动显示打开目录下的图像 #对不想要的图像，双击左键弹出一个确认对话框，选择yes将文件删除 global point1, point2, img, recimg,sw,sh,crop filelist=[]#图片文件列表 FileNum=0 pathname='' ObjectNum=len(ObjectName) Mflag=0#鼠标双功能切换标志 ObjectCode=0 #当前目标的类别代码 centerX=0 #标注矩形框中心的X坐标 centerY=0 #标注矩形框中心的坐标 boxw=0 #标注矩形框的宽度 boxh=0 #标注矩形框的高度 LabelData=[] #每幅图像对应的yolov8的TXT表数据 ImageData=[] #每幅标注的图像数据 BoxData=[]#每幅标注的矩形框数据表 CurImageNum=-1 #当前标定的图像序号 DrawRectangleFlag=False #画矩形标志 def InsertTable(data): global std2 str1 = "ClassID:" + str(data[0]) str2 = " xc:" + str(data[1]) str3 = " yc:" + str(data[2]) str4 = " w:"+str(data[3]) str5 = " h:"+str(data[4]) bbox = str1 + str2 + str3 + str4 + str5 std2.insert(tk.END, bbox) std2.insert(tk.END, '\r\n') return def ClearAnn(): global ImageData, LabelData, BoxData, CurImageNum, DrawRectangleFlag global Mflag, ObjectCode, centerX, centerY, boxw, boxh,std2 ImageData.clear() LabelData.clear() BoxData.clear() Mflag = 0 # 鼠标双功能切换标志 ObjectCode = 0 # 当前目标的类别代码 centerX = 0 # 标注矩形框中心的X坐标 centerY = 0 # 标注矩形框中心的坐标 boxw = 0 # 标注矩形框的宽度 boxh = 0 # 标注矩形框的高度 CurImageNum = -1 # 当前标定的图像序号 DrawRectangleFlag = False # 画矩形标志 std2.delete(0.0,tk.END) return #crop 剪裁窗口的鼠标响应 #双击鼠标左键，用剪裁出来的 def on_mouse2(event, x, y, flags, param): global crop,filelist,FileLength if event == cv2.EVENT_RBUTTONDBLCLK: file1 = file.asksaveasfilename(title='保存文件名字', initialdir="\image", filetypes=[('jpg文件', '*.jpg')]) if file1 == "": return cv2.imwrite(file1+'.jpg', crop) cv2.destroyWindow("crop") filelist = os.listdir(pathname) FileLength = len(filelist) elif event == cv2.EVENT_LBUTTONDBLCLK: fstr = pathname + '/' + filelist[FileNum] cv2.imwrite(fstr, crop) cv2.destroyWindow("crop") #批量标注，就是对某一目录下的同类图片给予一个标注号，首先通过移动滚动条选择图像目标的标注序号 #然后点击“批量标注”按钮，弹出批量标注图像文件的初始目录，可以浏览选择批量标注的新目录 #确定目录后就对整个目录下的图片打“标签”-就是设置为同一序号 def BatchTag(): global SamData, SamTag, ObjectCode, pathname,filelist if ObjectCode is None: mes.showwarning("无标签警告","没有设置标签") return pathname = file.askdirectory(title='批量标注图像文件目录', initialdir='\LhxArt\KerasCnn') if len(pathname) == 0: mes.showwarning("目录选择无效", "没有正确选择目录") return filelist = os.listdir(pathname) FileLength = len(filelist) fstr = pathname + '/' for i in range(FileLength): I1=cv2.imread(fstr+filelist[i]) if np.any(I1): I2=cv2.resize(I1,(ph,pw)) if np.any(I2): SamData.append(I2) SamTag.append(ObjectCode) CountStr.set(str(len(SamData))) ObjectCode=None return def on_ax_click(event): global ax, point1, point2, crop, FileLength, FileNum, pathname,filelist,yimg global CurImageNum, Mflag, centerX, centerY, boxw, boxh, DrawRectangleFlag,std2 if Mflag==1: if event.dblclick and event.button == 1: ans = mes.askyesno("图像删除确认", "确实想删除此图片吗？") if ans: fstr = pathname + '/' + filelist[FileNum] os.remove(fstr) filelist = os.listdir(pathname) FileLength = len(filelist) if FileNum > 0: FileNum -= 1 return elif event.button == 1: point1 = np.array([event.x, WH - event.y]) point2 = point1.copy() return elif event.dblclick and event.button==3: if point1[0] == point2[0] or point1[1] == point2[1]: return h,w=yimg.shape[:2] ymax = int(max(point1[1], point2[1])*(h/WH)) xmax = int(max(point1[0], point2[0])*(w/WW)) xmin = int(min(point1[0], point2[0])*(w/WW)) ymin = int(min(point1[1], point2[1])*(h/WH)) crop = yimg[ymin:ymax, xmin:xmax, :] cv2.imshow('crop', crop) cv2.setMouseCallback('crop', on_cv_mouse2) return elif event.button==3: if DrawRectangleFlag: centerX = np.around((point1[0] + point2[0]) / (2.0 * WW), decimals=5) centerY = np.around((point1[1] + point2[1]) / (2.0 * WH), decimals=5) boxw = np.around(np.abs(point2[0] - point1[0]) / WW, decimals=5) boxh = np.around(np.abs(point2[1] - point1[1]) / WH, decimals=5) DrawRectangleFlag = False Mflag = 2 return elif Mflag==2: if event.button==1: # 左键点击确定点BOX ls = [ObjectCode, centerX, centerY,boxw,boxh] if CurImageNum== FileNum: InsertTable(ls) BoxData.append(ls.copy()) else: fstr = pathname + '/' + filelist[FileNum] std2.insert(tk.END, fstr) std2.insert(tk.END, '\r\n') InsertTable(ls) ImageData.append(img) if len(ImageData)==1: but7.config(state=tk.ACTIVE) BoxData.append(ls.copy()) else: LabelData.append(BoxData.copy()) BoxData.clear() BoxData.append(ls.copy()) CurImageNum = FileNum Mflag = 1 return def on_ax_motion(event): global point1, point2, ax, img, recimg, GI, recimg, newimg,DrawRectangleFlag if event.button == 1: point2 = np.array([event.x, WH - event.y]) cv2.rectangle(recimg, point1, point2, (0, 255, 0), 2) newimg = cv2.bitwise_xor(np.uint8(img), np.uint8(recimg)) ax.imshow(newimg) recimg = np.zeros((WH, WW, 3), dtype=np.uint8) draw_set.draw() OutputFrame.update() DrawRectangleFlag=True return def on_ax_scroll(event): global img, recimg,filelist, FileNum, FileLength, pathname, ax, yimg, point1, newimg, ObjectCode, Mflag if Mflag==1: if pathname == "": return if event.button=="down" : if FileNum > 0: FileNum -= 1 else: if FileNum < FileLength - 1: FileNum += 1 fstr = pathname + '/' + filelist[FileNum] I = cv2.imread(fstr) imgsize = np.shape(I) # 图像尺寸大于1024，进行适当缩放 if imgsize[0] > 1024 or imgsize[1] > 1024: sv = 1024.0 / max(imgsize[0], imgsize[1]) img = cv2.resize(I, (int(sv * imgsize[1]), int(sv * imgsize[0]))) else: img = I yimg=img.copy() img=cv2.resize(img,(WW,WH)) img=cv2.cvtColor(img,cv2.COLOR_BGR2RGB) fig.clear() ax = fig.add_subplot(1, 1, 1, position=[0, 0, 1, 1]) ax.axis("off") ax.imshow(img) draw_set.draw() OutputFrame.update() elif Mflag==2: if event.button=="up" : if ObjectCode > 0: ObjectCode -= 1 else: if ObjectCode < ObjectNum - 1: ObjectCode += 1 drawChinese(newimg, point1, ObjectCode) return #ObjectName =['人','电脑','鼠标', '键盘', '订书机', '手机', '书', '台灯','杯子','笔'] #ObjectName = ['行人','汽车', '树木', '摩托', '交通灯', '狗', '路灯', '停车场','自行车','鲜花'] ObjectNum=len(ObjectName) def drawChinese(img, point, num): global ObjectName image = cv2.cvtColor(img, cv2.COLOR_BGR2RGB) # 将OpenCV图像转换为PIL图像 pil_image = Image.fromarray(image) # 准备写中文的工具 draw = ImageDraw.Draw(pil_image) font = ImageFont.truetype('simsun.ttc', 30) # 'simsun.ttc' 是常见的中文字体 # 写入中文文本 draw.text(point, ObjectName[num], font=font, fill=(255, 0, 0)) # 将PIL图像转换回OpenCV图像 img = cv2.cvtColor(np.array(pil_image), cv2.COLOR_RGB2BGR) ax.imshow(img) draw_set.draw() OutputFrame.update() return #crop 剪裁窗口的鼠标响应 #cv2窗口双击鼠标左键，用剪裁出来的 def on_cv_mouse2(event, x, y, flags, param): global crop,filelist,FileLength if event == cv2.EVENT_RBUTTONDBLCLK: file1 = file.asksaveasfilename(title='保存文件名字', initialdir="\image", filetypes=[('jpg文件', '*.jpg')]) if file1 == "": return cv2.imwrite(file1+'.jpg', crop) cv2.destroyWindow("crop") filelist = os.listdir(pathname) FileLength = len(filelist) elif event == cv2.EVENT_LBUTTONDBLCLK: fstr = pathname + '/' + filelist[FileNum] cv2.imwrite(fstr, crop) cv2.destroyWindow("crop") ######################################### def Yolov8Detect(): global ObjectName,model8 cap = cv2.VideoCapture('MyVideo.mp4') #cap.open(0, cv2.CAP_DSHOW) #model8.load(r"E:\LhxAgent\runs\detect\train35\weights\best.pt") cv2.waitKey(10) while True: ret, frame = cap.read() # 从本地视频通道采集一帧图像 img = cv2.resize(frame, (640, 480)) # 改变帧的长和宽为1024X800 results = model8.predict(img, show=False, save=False, verbose=False) boxes = results[0].boxes.xywh.cpu() indx = results[0].boxes.cls conf = results[0].boxes.conf class_names = [ObjectName[int(cls)] for cls in indx] vconf = [float(c) for c in conf] for box, name, score in zip(boxes, class_names,vconf): x_center, y_center, width, height = box.tolist() x1 = int(x_center - width / 2) y1 = int(y_center - height / 2) width = int(width) height = int(height) cv2.rectangle(img, (x1, y1), (x1 + width, y1 + height), (255, 0, 0), 2) pil_image = Image.fromarray(img) draw = ImageDraw.Draw(pil_image) font = ImageFont.truetype('simsun.ttc', 30) # 'simsun.ttc' 是常见的中文字体 # 写入中文文本 str1 = name + " " + str(score)[0:4] draw.text((x1, y1), str1, font=font, fill=(255, 0, 0)) # 将PIL图像转换回OpenCV图像 #img = cv2.cvtColor(np.array(pil_image), cv2.COLOR_RGB2BGR) img=np.array(pil_image) cv2.imshow("img", img) k = cv2.waitKey(10) & 0xff if k == 27: # press 'ESC' to quit break cap.release() cv2.destroyAllWindows() return def Yolov8SingleDetect(): global GI, ObjectName, model8 #results = model8.predict(GI, show=False, save=False, verbose=False) results = model8.predict(GI, show=False, save=False, verbose=False) img =GI # results[0].plot() boxes = results[0].boxes.xywh.cpu() #img = results[0].plot() #获得检测目标框的类别-是张量 indx=results[0].boxes.cls conf = results[0].boxes.conf class_names = [ObjectName[int(cls)] for cls in indx] vconf = [float(c) for c in conf] print(vconf) #print("result=",class_names) for box, idx,score in zip(boxes, class_names,vconf): x_center, y_center, width, height = box.tolist() x1 = int(x_center - width / 2) y1 = int(y_center - height / 2) width=int(width) height=int(height) cv2.imshow("img",GI) #print(name) cv2.rectangle(img, (x1, y1), (x1 + width, y1 + height), (255, 0, 0), 2) pil_image = Image.fromarray(img) draw = ImageDraw.Draw(pil_image) font = ImageFont.truetype('simsun.ttc', 30) # 'simsun.ttc' 是常见的中文字体 # 写入中文文本 str1=idx+" "+str(score)[0:4] draw.text((x1, y1), str1, font=font, fill=(255, 0, 0)) # 将PIL图像转换回OpenCV图像 img =np.array(pil_image)# cv2.cvtColor(np.array(pil_image), cv2.COLOR_RGB2BGR) cv2.imshow("img", img) return def TrainYoloV8(): global model8 model8 = YOLO("yolov8n.yaml") # build a new model from scratch model8 = YOLO("yolov8n.pt") model8.train(data="lihexi1.yaml", epochs=100, device="CPU", save=True) return def LoadModel(): global model8 #model8=YOLO(r'E:\LhxAgent\runs\detect\train2\weights\best.pt') model8 = YOLO(r'E:\LhxAgent\runs\detect\train2\weights\yolov8n.pt') return #========函数区结束=========== #=========菜单区============= menubar = tk.Menu(MainWin) # file menu fmenu = tk.Menu(menubar) fmenu.add_command(label='装入图像', command=LoadImage) fmenu.add_command(label='保存图像', command=SaveImage) fmenu.add_command(label='批量保存', command=callback1) fmenu.add_command(label='备用', command=callback1) # Image processing menu pmenu = tk.Menu(menubar) pmenu.add_command(label='开始爬取', command=StartCrawling) pmenu.add_command(label='暂停爬取', command=Suspending) pmenu.add_command(label='继续爬取', command=Resuming) pmenu.add_command(label='停止爬取', command=StopCrawling) # machine learning qmenu = tk.Menu(menubar) qmenu.add_command(label='图像清洗', command=CleanImage) qmenu.add_command(label='图像浏览', command=ScanImage) qmenu.add_command(label='批量换名', command=RenameImage) qmenu.add_command(label='转换成NPZ', command=NPZfile) lmenu = tk.Menu(menubar) lmenu.add_command(label='图像浏览', command=ScanImage) lmenu.add_command(label='标注类别名称', command=LoadClassName) lmenu.add_command(label='批量标注', command=LabelBatch) lmenu.add_command(label='标注结束', command=LabelEnd) lmenu.add_command(label='保存标注', command=LabelSave) tmenu = tk.Menu(menubar) tmenu.add_command(label='训练配置', command=TrainSetting) tmenu.add_command(label='训练启动', command=TrainStarting) tmenu.add_command(label='结果指标', command=TrainResultShow) tmenu.add_command(label='保存结果', command=SaveTrainResult) emenu = tk.Menu(menubar) emenu.add_command(label='打开图像', command=SelectImage) emenu.add_command(label='单幅测试', command=SingleImageTest) emenu.add_command(label='跟踪测试', command=ObjectTracking) emenu.add_command(label='保存测试结果', command=SaveTestResult) hmenu = tk.Menu(menubar) hmenu.add_command(label='图像爬取操作说明', command=HelpPicCrawl) hmenu.add_command(label='图像区域标注说明', command=HelpBatchAnn) hmenu.add_command(label='批量标注说明', command=HelpBatchAnn) hmenu.add_command(label='模型训练说明', command=callback1) menubar.add_cascade(label="文件操作", menu=fmenu) menubar.add_cascade(label="目标图像爬取", menu=pmenu) menubar.add_cascade(label="目标图像清洗", menu=qmenu) menubar.add_cascade(label="目标图像标注", menu=lmenu) menubar.add_cascade(label="目标图像训练", menu=tmenu) menubar.add_cascade(label="目标图像测试", menu=emenu) menubar.add_cascade(label="操作说明", menu=hmenu) MainWin.config(menu=menubar) #设置4个Frame 区， w1=600 h1=500 WW=550 WH=400 InputFrame =tk.Frame(MainWin,height =h1, width=w1) OutputFrame =tk.Frame(MainWin,height =h1, width=w1) butFrame = tk.Frame(MainWin,height=h1, width=w1) DataFrame = tk.Frame(MainWin,height=h1, width=w1) InputFrame.grid(row=0,column=0) OutputFrame.grid(row=0,column=1) butFrame.grid(row=1,column=0) DataFrame.grid(row=1,column=1,sticky=tk.N) #InputFrame Lab1=tk.Label(InputFrame,text='在此输入爬取图片主题词:',font=('Arial', 12),width=20, height=1) Lab1.grid(row=0,column=0,padx=10,pady=20) KeyStr=tk.StringVar() entry1=tk.Entry(InputFrame,font=('Arial', 12), width=20,textvariable=KeyStr) entry1.grid(row=0,column=1) KeyStr.set('') fig2 = plt.Figure(figsize=(WW/100, WH/100), dpi=100) # 设置空画布窗口，figsize为大小（英寸），dpi为分辨率 draw_set2 = FigureCanvasTkAgg(fig2, master=InputFrame) # 将空画布设置在tkinter的输出容器OutputFrame上 draw_set2.get_tk_widget().grid(row=1, column=0,columnspan=2) ax2 = fig2.add_subplot(1, 1, 1, position=[0, 0, 1, 1]) logo=cv2.imread('ImageFile/AnnLogo.jpg') logo=cv2.cvtColor(logo,cv2.COLOR_BGR2RGB) ax2.axis("off") ax2.imshow(logo) draw_set2.draw() #LogoImage = tk.PhotoImage(file='ImageFile/AnnLogo.png') #Lab2=tk.Label(InputFrame, image=LogoImage) #Lab2.grid(row=1,column=0,columnspan=2,padx=40) ########输出窗口 Lab3 = tk.Label(OutputFrame, text='输出窗口', font=('Arial', 14), width=16, height=1) Lab3.grid(row=0, column=0, pady=20) fig = plt.Figure(figsize=(WW/100, WH/100), dpi=100) # 设置空画布窗口，figsize为大小（英寸），dpi为分辨率 draw_set = FigureCanvasTkAgg(fig, master=OutputFrame) # 将空画布设置在tkinter的输出容器OutputFrame上 draw_set.get_tk_widget().grid(row=1, column=0) ax = fig.add_subplot(1, 1, 1, position=[0, 0, 1, 1]) draw_set.draw() ############################################ Target=[('0',0),('1',1),('2',2),('3',3),('4',4),('5',5),('6',6),('7',7),('8',8),('9',9)] TargetVector=np.zeros(10) TargetV=tk.IntVar() Target_startx=50 Target_starty=20 for txt,num in Target: rbut=tk.Radiobutton(butFrame, text=txt, value=num,font=('Arial', 12), width=3, height=1,command=SetTarget, variable=TargetV) rbut.place(x=Target_startx + num * 50, y=Target_starty) but1=tk.Button(butFrame, text='开始爬取', font=('Arial', 12), width=10, height=1,command=StartCrawling) but2=tk.Button(butFrame, text='停止爬取', font=('Arial', 12), width=10, height=1,command=StopCrawling) but3=tk.Button(butFrame, text='浏览图片', font=('Arial', 12), width=10, height=1,command=ScanImage) but4=tk.Button(butFrame, text='模型训练', font=('Arial', 12), width=10, height=1, command=TrainStarting) but5=tk.Button(butFrame, text='装入模型', font=('Arial', 12), width=10, height=1, command=LoadModel) but6=tk.Button(butFrame, text='方框标注', font=('Arial', 12), width=10, height=1, command=LabelBatch) but7=tk.Button(butFrame, text='结束标注', font=('Arial', 12), width=10, height=1, command=LabelEnd) but8=tk.Button(butFrame, text='标注存盘', font=('Arial', 12), width=10, height=1, command=LabelSave) but9=tk.Button(butFrame, text='目标识别', font=('Arial', 12), width=10, height=1, command=Yolov8Detect) but1.place(x=50,y=80) but2.place(x=250,y=80) but3.place(x=450,y=80) but4.place(x=50,y=120) but5.place(x=250,y=120) but6.place(x=450,y=120) but7.place(x=50,y=160) but8.place(x=250,y=160) but9.place(x=450,y=160) but7.config(state=tk.DISABLED) but8.config(state=tk.DISABLED) #Data Frame Lab4=tk.Label(DataFrame,text='网页计数:',font=('Arial', 12),width=10, height=1) Lab4.grid(row=0,column=0,pady=40) Lab5=tk.Label(DataFrame,text='文件名称:',font=('Arial', 12),width=10, height=1) Lab5.grid(row=1,column=0) Lab6=tk.Label(DataFrame,text='下载地址',font=('Arial', 12),width=10, height=1) Lab6.grid(row=2,column=0,pady=40) CountStr=tk.StringVar() entry4=tk.Entry(DataFrame,font=('Arial', 12),width=15, textvariable=CountStr) entry4.grid(row=0,column=1,pady=40) CountStr.set("0") FileStr=tk.StringVar() entry5=tk.Entry(DataFrame,font=('Arial', 12), width=30,textvariable=FileStr) entry5.grid(row=1,column=1) FileStr.set("") DownAddrStr=tk.StringVar() entry6=tk.Entry(DataFrame,font=('Arial', 12), width=50,textvariable=DownAddrStr) entry6.grid(row=2,column=1,pady=40) #Thd1=thread.Thread(target=CrawlPicture,args=(300,500,)) #Thd1.setDaemon(True) #Thd1.start() MainWin.mainloop()怎么使用