前言

dlib提供了dlib.correlation_tracker()類用于跟蹤目標。
官方文檔入口：http://dlib.net/python/index.html#dlib.correlation_tracker
不復(fù)雜，就不介紹了，后面會直接給出兩個程序，有注釋。

程序1

# -*- coding: utf-8 -*- import sys import dlib import cv2tracker = dlib.correlation_tracker() # 導(dǎo)入correlation_tracker()類 cap = cv2.VideoCapture(0) # OpenCV打開攝像頭 start_flag = True # 標記，是否是第一幀，若在第一幀需要先初始化 selection = None # 實時跟蹤鼠標的跟蹤區(qū)域 track_window = None # 要檢測的物體所在區(qū)域 drag_start = None # 標記，是否開始拖動鼠標# 鼠標點擊事件回調(diào)函數(shù) def onMouseClicked(event, x, y, flags, param):global selection, track_window, drag_start # 定義全局變量if event == cv2.EVENT_LBUTTONDOWN: # 鼠標左鍵按下drag_start = (x, y)track_window = Noneif drag_start: # 是否開始拖動鼠標，記錄鼠標位置xMin = min(x, drag_start[0])yMin = min(y, drag_start[1])xMax = max(x, drag_start[0])yMax = max(y, drag_start[1])selection = (xMin, yMin, xMax, yMax)if event == cv2.EVENT_LBUTTONUP: # 鼠標左鍵松開drag_start = Nonetrack_window = selectionselection = Noneif __name__ == '__main__':cv2.namedWindow("image", cv2.WINDOW_AUTOSIZE)cv2.setMouseCallback("image", onMouseClicked)# opencv的bgr格式圖片轉(zhuǎn)換成rgb格式# b, g, r = cv2.split(frame)# frame2 = cv2.merge([r, g, b])while(1):ret, frame = cap.read() # 從攝像頭讀入1幀if start_flag == True: # 如果是第一幀，需要先初始化# 這里是初始化，窗口中會停在當前幀，用鼠標拖拽一個框來指定區(qū)域，隨后會跟蹤這個目標；我們需要先找到目標才能跟蹤不是嗎？while True:img_first = frame.copy() # 不改變原來的幀，拷貝一個新的出來if track_window: # 跟蹤目標的窗口畫出來了，就實時標出來cv2.rectangle(img_first, (track_window[0], track_window[1]), (track_window[2], track_window[3]), (0,0,255), 1)elif selection: # 跟蹤目標的窗口隨鼠標拖動實時顯示cv2.rectangle(img_first, (selection[0], selection[1]), (selection[2], selection[3]), (0,0,255), 1)cv2.imshow("image", img_first)# 按下回車，退出循環(huán)if cv2.waitKey(5) == 13:breakstart_flag = False # 初始化完畢，不再是第一幀了tracker.start_track(frame, dlib.rectangle(track_window[0], track_window[1], track_window[2], track_window[3])) # 跟蹤目標，目標就是選定目標窗口中的else:tracker.update(frame) # 更新，實時跟蹤box_predict = tracker.get_position() # 得到目標的位置cv2.rectangle(frame,(int(box_predict.left()),int(box_predict.top())),(int(box_predict.right()),int(box_predict.bottom())),(0,255,255),1) # 用矩形框標注出來cv2.imshow("image", frame)# 如果按下ESC鍵，就退出if cv2.waitKey(10) == 27:breakcap.release()cv2.destroyAllWindows()

注：如果程序卡了，就調(diào)一下cv2.waitKey()中的參數(shù)，也就是延時時間，調(diào)小即可。

運行結(jié)果

初始時，窗口中只會顯示第一幀的圖像；
使用鼠標拖拽一個框，紅框中目標后，按回車，設(shè)置框內(nèi)為識別目標；
實時識別，以橙框標出；
按ESC鍵退出。

（csdn只能上傳2M的圖片，真心難受）

程序2

由于前面那個程序，只是熟悉下函數(shù)寫的，我覺得用起來蛋疼，所以又重新封裝了一下。看起來舒服多了。

# -*- coding: utf-8 -*- import sys import dlib import cv2class myCorrelationTracker(object):def __init__(self, windowName='default window', cameraNum=0):# 自定義幾個狀態(tài)標志self.STATUS_RUN_WITHOUT_TRACKER = 0 # 不跟蹤目標，但是實時顯示self.STATUS_RUN_WITH_TRACKER = 1 # 跟蹤目標，實時顯示self.STATUS_PAUSE = 2 # 暫停，卡在當前幀self.STATUS_BREAK = 3 # 退出self.status = self.STATUS_RUN_WITHOUT_TRACKER # 指示狀態(tài)的變量# 這幾個跟前面程序1定義的變量一樣self.track_window = None # 實時跟蹤鼠標的跟蹤區(qū)域self.drag_start = None # 要檢測的物體所在區(qū)域self.start_flag = True # 標記，是否開始拖動鼠標# 創(chuàng)建好顯示窗口cv2.namedWindow(windowName, cv2.WINDOW_AUTOSIZE)cv2.setMouseCallback(windowName, self.onMouseClicked)self.windowName = windowName# 打開攝像頭self.cap = cv2.VideoCapture(cameraNum)# correlation_tracker()類，跟蹤器，跟程序1中一樣self.tracker = dlib.correlation_tracker()# 當前幀self.frame = None# 按鍵處理函數(shù)def keyEventHandler(self):keyValue = cv2.waitKey(5) # 每隔5ms讀取一次按鍵的鍵值if keyValue == 27: # ESCself.status = self.STATUS_BREAKif keyValue == 32: # 空格if self.status != self.STATUS_PAUSE: # 按下空格，暫停播放，可以選定跟蹤的區(qū)域#print self.statusself.status = self.STATUS_PAUSE#print self.statuselse: # 再按次空格，重新播放，但是不進行目標識別if self.track_window:self.status = self.STATUS_RUN_WITH_TRACKERself.start_flag = Trueelse:self.status = self.STATUS_RUN_WITHOUT_TRACKERif keyValue == 13: # 回車#print '**'if self.status == self.STATUS_PAUSE: # 按下空格之后if self.track_window: # 如果選定了區(qū)域，再按回車，表示確定選定區(qū)域為跟蹤目標self.status = self.STATUS_RUN_WITH_TRACKERself.start_flag = True# 任務(wù)處理函數(shù) def processHandler(self):# 不跟蹤目標，但是實時顯示if self.status == self.STATUS_RUN_WITHOUT_TRACKER:ret, self.frame = self.cap.read()cv2.imshow(self.windowName, self.frame)# 暫停，暫停時使用鼠標拖動紅框，選擇目標區(qū)域，與程序1類似elif self.status == self.STATUS_PAUSE:img_first = self.frame.copy() # 不改變原來的幀，拷貝一個新的變量出來if self.track_window: # 跟蹤目標的窗口畫出來了，就實時標出來cv2.rectangle(img_first, (self.track_window[0], self.track_window[1]), (self.track_window[2], self.track_window[3]), (0,0,255), 1)elif self.selection: # 跟蹤目標的窗口隨鼠標拖動實時顯示cv2.rectangle(img_first, (self.selection[0], self.selection[1]), (self.selection[2], self.selection[3]), (0,0,255), 1)cv2.imshow(self.windowName, img_first)# 退出elif self.status == self.STATUS_BREAK:self.cap.release() # 釋放攝像頭cv2.destroyAllWindows() # 釋放窗口sys.exit() # 退出程序# 跟蹤目標，實時顯示elif self.status == self.STATUS_RUN_WITH_TRACKER:ret, self.frame = self.cap.read() # 從攝像頭讀取一幀if self.start_flag: # 如果是第一幀，需要先初始化self.tracker.start_track(self.frame, dlib.rectangle(self.track_window[0], self.track_window[1], self.track_window[2], self.track_window[3])) # 開始跟蹤目標self.start_flag = False # 不再是第一幀else:self.tracker.update(self.frame) # 更新# 得到目標的位置，并顯示box_predict = self.tracker.get_position() cv2.rectangle(self.frame,(int(box_predict.left()),int(box_predict.top())),(int(box_predict.right()),int(box_predict.bottom())),(0,255,255),1)cv2.imshow(self.windowName, self.frame)# 鼠標點擊事件回調(diào)函數(shù)def onMouseClicked(self, event, x, y, flags, param):if event == cv2.EVENT_LBUTTONDOWN: # 鼠標左鍵按下self.drag_start = (x, y)self.track_window = Noneif self.drag_start: # 是否開始拖動鼠標，記錄鼠標位置xMin = min(x, self.drag_start[0])yMin = min(y, self.drag_start[1])xMax = max(x, self.drag_start[0])yMax = max(y, self.drag_start[1])self.selection = (xMin, yMin, xMax, yMax)if event == cv2.EVENT_LBUTTONUP: # 鼠標左鍵松開self.drag_start = Noneself.track_window = self.selectionself.selection = Nonedef run(self):while(1):self.keyEventHandler()self.processHandler()if __name__ == '__main__':testTracker = myCorrelationTracker(windowName='image', cameraNum=1)testTracker.run()

注：如果程序卡了，就調(diào)一下cv2.waitKey()中的參數(shù)，也就是延時時間，調(diào)小即可。

運行結(jié)果

操作有一些改變：
初始時，會自動從攝像頭采集圖像顯示；
按下空格，暫停；此時若再按空格，恢復(fù)實時顯示，但不進行目標跟蹤；
暫停時，拖動鼠標會顯示紅框，按下回車，將紅框內(nèi)物體視為目標進行識別；
隨后實時識別，以橙框標出；
按ESC鍵退出。

官方例程

#!/usr/bin/python # The contents of this file are in the public domain. See LICENSE_FOR_EXAMPLE_PROGRAMS.txt # # This example shows how to use the correlation_tracker from the dlib Python # library. This object lets you track the position of an object as it moves # from frame to frame in a video sequence. To use it, you give the # correlation_tracker the bounding box of the object you want to track in the # current video frame. Then it will identify the location of the object in # subsequent frames. # # In this particular example, we are going to run on the # video sequence that comes with dlib, which can be found in the # examples/video_frames folder. This video shows a juice box sitting on a table # and someone is waving the camera around. The task is to track the position of # the juice box as the camera moves around. # # # COMPILING/INSTALLING THE DLIB PYTHON INTERFACE # You can install dlib using the command: # pip install dlib # # Alternatively, if you want to compile dlib yourself then go into the dlib # root folder and run: # python setup.py install # or # python setup.py install --yes USE_AVX_INSTRUCTIONS # if you have a CPU that supports AVX instructions, since this makes some # things run faster. # # Compiling dlib should work on any operating system so long as you have # CMake and boost-python installed. On Ubuntu, this can be done easily by # running the command: # sudo apt-get install libboost-python-dev cmake # # Also note that this example requires scikit-image which can be installed # via the command: # pip install scikit-image # Or downloaded from http://scikit-image.org/download.html. import os import globimport dlib from skimage import io# Path to the video frames video_folder = os.path.join("..", "examples", "video_frames")# Create the correlation tracker - the object needs to be initialized # before it can be used tracker = dlib.correlation_tracker()win = dlib.image_window() # We will track the frames as we load them off of disk for k, f in enumerate(sorted(glob.glob(os.path.join(video_folder, "*.jpg")))):print("Processing Frame {}".format(k))img = io.imread(f)# We need to initialize the tracker on the first frameif k == 0:# Start a track on the juice box. If you look at the first frame you# will see that the juice box is contained within the bounding# box (74, 67, 112, 153).tracker.start_track(img, dlib.rectangle(74, 67, 112, 153))else:# Else we just attempt to track from the previous frametracker.update(img)win.clear_overlay()win.set_image(img)win.add_overlay(tracker.get_position())dlib.hit_enter_to_continue()

吐槽：
已經(jīng)寫了四篇有關(guān)dlib的學(xué)習筆記了。dlib這個庫的確很方便，能夠很輕松地讓我們實現(xiàn)一些基礎(chǔ)的識別任務(wù)，比如人臉識別。但是如果想要在各種識別任務(wù)中有更好的效果，肯定是不能只用他給的模型的。那也就是說需要自己訓(xùn)練了，看到官方文檔中也提供了一些訓(xùn)練以及自己構(gòu)建神經(jīng)網(wǎng)絡(luò)等的api接口，下次有時間再來整理一下程序。最近學(xué)校又是校運會，又要看數(shù)學(xué)（矩陣論、凸優(yōu)化），不過這樣抽時間出來寫寫程序心情也舒暢了不少。
ヽ(･ω･｡)ﾉ

總結(jié)

以上是生活随笔為你收集整理的python dlib学习（四）：单目标跟踪的全部內(nèi)容，希望文章能夠幫你解決所遇到的問題。

如果覺得生活随笔網(wǎng)站內(nèi)容還不錯，歡迎將生活随笔推薦給好友。