import datetime

class FPS:
?? ?def __init__(self):
?? ??? ?# 定義開始時(shí)間、結(jié)束時(shí)間和總幀數(shù)
?? ??? ?self._start = None
?? ??? ?self._end = None
?? ??? ?self._numFrames = 0

?? ?def start(self):
?? ??? ?# 開始計(jì)時(shí)
?? ??? ?self._start = datetime.datetime.now()
?? ??? ?return self

?? ?def stop(self):
?? ??? ?# 停止計(jì)時(shí)
?? ??? ?self._end = datetime.datetime.now()

?? ?def update(self):
?? ??? ?# 增加在開始和結(jié)束間隔期間檢查的總幀數(shù)
?? ??? ?self._numFrames += 1

?? ?def elapsed(self):
?? ??? ?# 返回開始和結(jié)束間隔之間的總秒數(shù)
?? ??? ?return (self._end - self._start).total_seconds()

?? ?def fps(self):
?? ??? ?# 計(jì)算每秒幀數(shù)
?? ??? ?return self._numFrames / self.elapsed()

import cv2
from fps import FPS
def resize(image, width=None, height=None, inter=cv2.INTER_AREA):
    # 初始化要調(diào)整大小的圖像的尺寸并抓取圖像大小
    dim = None
    (h, w) = image.shape[:2]
    # 如果寬高都為None，則返回原圖
    if width is None and height is None:
        return image
    # 檢查寬度是否為None
    if width is None:
        # 計(jì)算高度的比例并構(gòu)造尺寸
        r = height / float(h)
        dim = (int(w * r), height)
    # 否則，高度為 None
    else:
        # 計(jì)算寬度的比例并構(gòu)造尺寸
        r = width / float(w)
        dim = (width, int(h * r))
    resized = cv2.resize(image, dim, interpolation=inter)
    return resized

videos = 0
tracker_type = 'kcf'

# 提取 OpenCV 版本信息
(major, minor) = cv2.__version__.split(".")[:2]
# 如果我們使用 OpenCV 3.2 或之前版本，我們可以使用特殊的工廠函數(shù)來創(chuàng)建我們的對(duì)象跟蹤器
if int(major) == 3 and int(minor) < 3:
    tracker = cv2.Tracker_create(tracker_type)
# 否則，對(duì)于 OpenCV 3.3 或更新版本，我們需要顯式調(diào)用對(duì)應(yīng)的對(duì)象跟蹤器構(gòu)造函數(shù)：
else:
    # 初始化一個(gè)字典，將字符串映射到其對(duì)應(yīng)的 OpenCV 對(duì)象跟蹤器實(shí)現(xiàn)
    OPENCV_OBJECT_TRACKERS = {
        "csrt": cv2.TrackerCSRT_create,
        "kcf": cv2.TrackerKCF_create,
        "boosting": cv2.legacy.TrackerBoosting_create,
        "mil": cv2.TrackerMIL_create,
        "tld": cv2.legacy.TrackerTLD_create,
        "medianflow": cv2.legacy.TrackerMedianFlow_create,
        "mosse": cv2.legacy.TrackerMOSSE_create
    }
    # 使用我們的 OpenCV 對(duì)象跟蹤器對(duì)象字典獲取適當(dāng)?shù)膶?duì)象跟蹤器
    tracker = OPENCV_OBJECT_TRACKERS[tracker_type]()

# 初始化我們要追蹤的物體的邊界框坐標(biāo)
initBB = None
vs = cv2.VideoCapture(videos)
fps = None

# 循環(huán)播放視頻流中的幀
while True:
    # 抓取當(dāng)前幀。
    (grabbed, frame) = vs.read()
    if not grabbed:
        break
    # 調(diào)整框架大小并獲取框架尺寸。
    frame = resize(frame, width=500)
    (H, W) = frame.shape[:2]

    # 檢查是否正在跟蹤一個(gè)對(duì)象
    if initBB is not None:
        # 抓取物體的新邊界框坐標(biāo)
        (success, box) = tracker.update(frame)
        # 檢查跟蹤是否成功
        if success:
            (x, y, w, h) = [int(v) for v in box]
            cv2.rectangle(frame, (x, y), (x + w, y + h),
                          (0, 255, 0), 2)
        # 更新 FPS 計(jì)數(shù)器
        fps.update()
        fps.stop()
        # 初始化在框架上顯示的信息集
        info = [
            ("Tracker", tracker_type),
            ("Success", "Yes" if success else "No"),
            ("FPS", "{:.2f}".format(fps.fps())),
        ]
        # 遍歷信息元組并將它們繪制在框架上
        for (i, (k, v)) in enumerate(info):
            text = "{}: {}".format(k, v)
            cv2.putText(frame, text, (10, H - ((i * 20) + 20)),
                        cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0, 0, 255), 2)

    # 顯示輸出幀
    cv2.imshow("Frame", frame)
    key = cv2.waitKey(1) & 0xFF

   # 使用's'鍵選擇一個(gè)邊界框來跟蹤
    if key == ord("s"):
        # 選擇跟蹤的對(duì)象的邊界框（選擇 ROI 后按 ENTER 或 SPACE）
        initBB = cv2.selectROI("Frame", frame, fromCenter=False,
                               showCrosshair=True)
        # 使用提供的邊界框坐標(biāo)啟動(dòng) OpenCV 對(duì)象跟蹤器，然后也啟動(dòng) FPS 吞吐量估計(jì)器
        tracker.init(frame, initBB)
        fps = FPS().start()
    # 如果 `q` 鍵被按下，則退出循環(huán)
    elif key == ord("q"):
        break
vs.release()
cv2.destroyAllWindows()