import cv2
import numpy as np
import time
from concurrent.futures import ThreadPoolExecutor

conf_threshold = 0.5
nms_threshold = 0.4
Width = 416
Height = 416

net = cv2.dnn.readNet('../needFiles/yolov3.weights', '../needFiles/yolov3.cfg')
with open('../needFiles/coco.names', 'r') as f:
    classes = f.read().strip().split('\n')

color_list = np.random.uniform(0, 255, size=(len(classes), 3))

def fetch_frame(cap):
    ...
def process_frame(frame):
    ...

executor = ThreadPoolExecutor(max_workers=2)
frame_future = executor.submit(fetch_frame, cap)

while True:
    ...
    ret, frame, height, width, channels = frame_future.result()
    frame_future = executor.submit(fetch_frame, cap)
    processed_frame = executor.submit(process_frame, frame).result()
    ...

if cv2.waitKey(1) & 0xFF == ord('q'):
    break
...
cap.release()
cv2.destroyAllWindows()

# 導(dǎo)入必要的庫(kù)
import cv2
import numpy as np
import time
from concurrent.futures import ThreadPoolExecutor

# 設(shè)置置信度閾值和非極大值抑制（NMS）閾值
conf_threshold = 0.5
nms_threshold = 0.4
Width = 416
Height = 416

# 加載預(yù)訓(xùn)練的 YOLOv3 模型
net = cv2.dnn.readNet('../needFiles/yolov3.weights', '../needFiles/yolov3.cfg')

# 加載可識(shí)別的類(lèi)名
with open('../needFiles/coco.names', 'r') as f:
    classes = f.read().strip().split('\n')

# 為不同的類(lèi)別創(chuàng)建一個(gè)顏色列表
color_list = np.random.uniform(0, 255, size=(len(classes), 3))

# 打開(kāi)攝像頭進(jìn)行視頻幀的捕獲
cap = cv2.VideoCapture(0)

# 初始化一個(gè)ThreadPoolExecutor用于多線程處理
executor = ThreadPoolExecutor(max_workers=2)


# 定義fetch_frame函數(shù)，從視頻流中獲取視頻幀
def fetch_frame(cap):
    ret, frame = cap.read()  # 讀取一幀圖像
    height, width, channels = frame.shape  # 獲取圖像的尺寸和通道信息
    return ret, frame, height, width, channels


# 定義process_frame函數(shù)，處理每幀圖像并進(jìn)行目標(biāo)檢測(cè)
def process_frame(frame):
    # 將幀轉(zhuǎn)換為模型的輸入格式
    blob = cv2.dnn.blobFromImage(frame, 1 / 255.0, (416, 416), swapRB=True, crop=False)
    net.setInput(blob)
    output_layers = net.getUnconnectedOutLayersNames()  # 獲取輸出層的名字
    layer_outputs = net.forward(output_layers)  # 進(jìn)行前向傳播，獲取檢測(cè)結(jié)果

    boxes = []  # 用于存儲(chǔ)檢測(cè)到的邊界框
    confidences = []  # 用于存儲(chǔ)邊界框的置信度
    class_ids = []  # 用于存儲(chǔ)邊界框的類(lèi)別ID

    # 循環(huán)每個(gè)輸出層的檢測(cè)結(jié)果
    for output in layer_outputs:
        for detection in output:
            scores = detection[5:]  # 獲取類(lèi)別的得分
            class_id = np.argmax(scores)  # 獲取得分最高的類(lèi)別ID
            confidence = scores[class_id]  # 獲取得分最高的置信度

            # 過(guò)濾低置信度的檢測(cè)結(jié)果
            if confidence > conf_threshold:
                center_x = int(detection[0] * width)
                center_y = int(detection[1] * height)
                w = int(detection[2] * width)
                h = int(detection[3] * height)

                # 計(jì)算邊界框的位置和尺寸
                x = int(center_x - w / 2)
                y = int(center_y - h / 2)

                # 將邊界框的位置、尺寸、置信度和類(lèi)別ID添加到列表中
                boxes.append([x, y, w, h])
                confidences.append(float(confidence))
                class_ids.append(class_id)

    # 使用非極大值抑制去除重疊的邊界框
    indices = cv2.dnn.NMSBoxes(boxes, confidences, conf_threshold, nms_threshold)

    # 在原圖上繪制邊界框和類(lèi)別標(biāo)簽
    for i in indices.flatten():
        box = boxes[i]
        x = box[0]
        y = box[1]
        w = box[2]
        h = box[3]
        label = str(classes[class_ids[i]])
        color = color_list[class_ids[i]]
        cv2.rectangle(frame, (x, y), (x + w, y + h), color, 2)
        cv2.putText(frame, label, (x, y - 10), cv2.FONT_HERSHEY_SIMPLEX, 0.5, color, 2)

    return frame


# 在進(jìn)入循環(huán)前，先讀取一幀以開(kāi)始異步處理
frame_future = executor.submit(fetch_frame, cap)

# 主循環(huán)
while True:
    start = time.time()  # 記錄開(kāi)始處理的時(shí)間點(diǎn)
    # 獲取當(dāng)前幀和相應(yīng)信息
    ret, frame, height, width, channels = frame_future.result()

    # 異步讀取下一幀
    frame_future = executor.submit(fetch_frame, cap)

    # 如果當(dāng)前幀讀取成功，則繼續(xù)處理
    if ret:
        # 使用線程池異步處理當(dāng)前幀
        processed_frame = executor.submit(process_frame, frame).result()

        # 計(jì)算FPS
        end = time.time()
        fps = 1 / (end - start)
        # 在處理好的幀上顯示FPS
        cv2.putText(processed_frame, "FPS: " + str(round(fps, 2)), (10, 30), cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0, 255, 0),
                    2)
        # 顯示處理好的幀
        cv2.imshow('frame', processed_frame)

        # 如果我們按下 ‘q' 鍵，退出循環(huán)
        if cv2.waitKey(1) & 0xFF == ord('q'):
            break
    else:
        break  # 如果幀沒(méi)有被成功讀取，退出循環(huán)

# 釋放視頻捕獲對(duì)象和銷(xiāo)毀所有OpenCV窗口
cap.release()
cv2.destroyAllWindows()

   # 加載預(yù)訓(xùn)練的 YOLOv3 模型
   net = cv2.dnn.readNetFromDarknet('../needFiles/yolov3.cfg', '../needFiles/yolov3.weights')
   
   # 啟用 CUDA
   net.setPreferableBackend(cv2.dnn.DNN_BACKEND_CUDA)
   net.setPreferableTarget(cv2.dnn.DNN_TARGET_CUDA)