10分钟学会使用YOLO及Opencv实现目标检测
首先打开 yolo_video.py文件并插入以下代码:
# import the necessary packages import numpy as np import argparse import imutils import time import cv2 import os # construct the argument parse and parse the arguments ap = argparse.ArgumentParser() ap.add_argument("-i", "--input", required=True, help="path to input video") ap.add_argument("-o", "--output", required=True, help="path to output video") ap.add_argument("-y", "--yolo", required=True, help="base path to YOLO directory") ap.add_argument("-c", "--confidence", type=float, default=0.5, help="minimum probability to filter weak detections") ap.add_argument("-t", "--threshold", type=float, default=0.3, help="threshold when applyong non-maxima suppression") args = vars(ap.parse_args())
同样,首先从导入相关数据包和命令行参数开始。与之前不同的是,此脚本没有-- image参数,取而代之的是量个视频路径:
-- input :输入视频文件的路径;
-- output :输出视频文件的路径;
视频的输入可以是手机拍摄的短视频或者是网上搜索到的视频。另外,也可以通过将多张照片合成为一个短视频也可以。本博客使用的是在PyImageSearch上找到来自imutils的VideoStream类的 示例。
下面的代码与处理图形时候相同:
# load the COCO class labels our YOLO model was trained on labelsPath = os.path.sep.join([args["yolo"], "coco.names"]) LABELS = open(labelsPath).read().strip().split("\n") # initialize a list of colors to represent each possible class label np.random.seed(42) COLORS = np.random.randint(0, 255, size=(len(LABELS), 3), dtype="uint8") # derive the paths to the YOLO weights and model configuration weightsPath = os.path.sep.join([args["yolo"], "yolov3.weights"]) configPath = os.path.sep.join([args["yolo"], "yolov3.cfg"]) # load our YOLO object detector trained on COCO dataset (80 classes) # and determine only the *output* layer names that we need from YOLO print("[INFO] loading YOLO from disk...") net = cv2.dnn.readNetFromDarknet(configPath, weightsPath) ln = net.getLayerNames() ln = [ln[i[0] - 1] for i in net.getUnconnectedOutLayers()]
在这里,加载标签并生成相应的颜色,然后加载YOLO模型并确定输出层名称。
接下来,将处理一些特定于视频的任务:
# initialize the video stream, pointer to output video file, and # frame dimensions vs = cv2.VideoCapture(args["input"]) writer = None (W, H) = (None, None) # try to determine the total number of frames in the video file try: prop = cv2.cv.CV_CAP_PROP_frame_COUNT if imutils.is_cv2() \ else cv2.CAP_PROP_frame_COUNT total = int(vs.get(prop)) print("[INFO] {} total frames in video".format(total)) # an error occurred while trying to determine the total # number of frames in the video file except: print("[INFO] could not determine # of frames in video") print("[INFO] no approx. completion time can be provided") total = -1
在上述代码块中:
打开一个指向视频文件的文件指针,循环读取帧;
初始化视频编写器 (writer)和帧尺寸;
尝试确定视频文件中的总帧数(total),以便估计整个视频的处理时间;
之后逐个处理帧:
# loop over frames from the video file stream while True: # read the next frame from the file (grabbed, frame) = vs.read() # if the frame was not grabbed, then we have reached the end # of the stream if not grabbed: break # if the frame dimensions are empty, grab them if W is None or H is None: (H, W) = frame.shape[:2]
上述定义了一个 while循环, 然后从第一帧开始进行处理,并且会检查它是否是视频的最后一帧。接下来,如果尚未知道帧的尺寸,就会获取一下对应的尺寸。
接下来,使用当前帧作为输入执行YOLO的前向传递 :
ect Detection with OpenCVPython # construct a blob from the input frame and then perform a forward # pass of the YOLO object detector, giving us our bounding boxes # and associated probabilities blob = cv2.dnn.blobFromImage(frame, 1 / 255.0, (416, 416), swapRB=True, crop=False) net.setInput(blob) start = time.time() layerOutputs = net.forward(ln) end = time.time() # initialize our lists of detected bounding boxes, confidences, # and class IDs, respectively boxes = [] confidences = [] classIDs = []
在这里,构建一个 blob 并将其传递通过网络,从而获得预测。然后继续初始化之前在图像目标检测中使用过的三个列表: boxes 、 confidences、classIDs :
# loop over each of the layer outputs for output in layerOutputs: # loop over each of the detections for detection in output: # extract the class ID and confidence (i.e., probability) # of the current object detection scores = detection[5:] classID = np.argmax(scores) confidence = scores[classID] # filter out weak predictions by ensuring the detected # probability is greater than the minimum probability if confidence > args["confidence"]: # scale the bounding box coordinates back relative to # the size of the image, keeping in mind that YOLO # actually returns the center (x, y)-coordinates of # the bounding box followed by the boxes' width and # height box = detection[0:4] * np.array([W, H, W, H]) (centerX, centerY, width, height) = box.astype("int") # use the center (x, y)-coordinates to derive the top # and and left corner of the bounding box x = int(centerX - (width / 2)) y = int(centerY - (height / 2)) # update our list of bounding box coordinates, # confidences, and class IDs boxes.append([x, y, int(width), int(height)]) confidences.append(float(confidence)) classIDs.append(classID)
在上述代码中,与图像目标检测相同的有:
循环输出层和检测;
提取 classID并过滤掉弱预测;
计算边界框坐标;
更新各自的列表;
接下来,将应用非最大值抑制:
# apply non-maxima suppression to suppress weak, overlapping # bounding boxes idxs = cv2.dnn.NMSBoxes(boxes, confidences, args["confidence"], args["threshold"]) # ensure at least one detection exists if len(idxs) > 0: # loop over the indexes we are keeping for i in idxs.flatten(): # extract the bounding box coordinates (x, y) = (boxes[i][0], boxes[i][1]) (w, h) = (boxes[i][2], boxes[i][3]) # draw a bounding box rectangle and label on the frame color = [int(c) for c in COLORS[classIDs[i]]] cv2.rectangle(frame, (x, y), (x + w, y + h), color, 2) text = "{}: {:.4f}".format(LABELS[classIDs[i]], confidences[i]) cv2.putText(frame, text, (x, y - 5), cv2.FONT_HERSHEY_SIMPLEX, 0.5, color, 2)
同样的,在上述代码中与图像目标检测相同的有:
使用cv2.dnn.NMSBoxes函数用于抑制弱的重叠边界框,可以在此处阅读有关非最大值抑制的更多信息;
循环遍历由NMS计算的idx,并绘制相应的边界框+标签;
最终的部分代码如下:
# check if the video writer is None if writer is None: # initialize our video writer fourcc = cv2.VideoWriter_fourcc(*"MJPG") writer = cv2.VideoWriter(args["output"], fourcc, 30, (frame.shape[1], frame.shape[0]), True) # some information on processing single frame if total > 0: elap = (end - start) print("[INFO] single frame took {:.4f} seconds".format(elap)) print("[INFO] estimated total time to finish: {:.4f}".format( elap * total)) # write the output frame to disk writer.write(frame) # release the file pointers print("[INFO] cleaning up...") writer.release() vs.release()
总结一下:
初始化视频编写器(writer),一般在循环的第一次迭代被初始化;
打印出对处理视频所需时间的估计;
将帧(frame)写入输出视频文件;
清理和释放指针;
现在,打开一个终端并执行以下命令:
$ python yolo_video.py --input videos/car_chase_01.mp4 \ --output output/car_chase_01.avi --yolo yolo-coco [INFO] loading YOLO from disk... [INFO] 583 total frames in video [INFO] single frame took 0.3500 seconds [INFO] estimated total time to finish: 204.0238 [INFO] cleaning up...
在视频/ GIF中,你不仅可以看到被检测到的车辆,还可以检测到人员以及交通信号灯!
YOLO目标检测器在该视频中表现相当不错。让现在尝试同一车追逐视频中的不同视频:
$ python yolo_video.py --input videos/car_chase_02.mp4 \ --output output/car_chase_02.avi --yolo yolo-coco [INFO] loading YOLO from disk... [INFO] 3132 total frames in video [INFO] single frame took 0.3455 seconds [INFO] estimated total time to finish: 1082.0806 [INFO] cleaning up...
YOLO再一次能够检测到行人!或者嫌疑人回到他们的车中并继续追逐:
$ python yolo_video.py --input videos/car_chase_03.mp4 \ --output output/car_chase_03.avi --yolo yolo-coco [INFO] loading YOLO from disk... [INFO] 749 total frames in video [INFO] single frame took 0.3442 seconds [INFO] estimated total time to finish: 257.8418 [INFO] cleaning up...
最后一个例子,让我们看看如何使用YOLO作为构建流量计数器:
$ python yolo_video.py --input videos/overpass.mp4 \ --output output/overpass.avi --yolo yolo-coco [INFO] loading YOLO from disk... [INFO] 812 total frames in video [INFO] single frame took 0.3534 seconds [INFO] estimated total time to finish: 286.9583 [INFO] cleaning up...
下面汇总YOLO视频对象检测完整视频:
Quaker Oats汽车追逐视频;
Vlad Kiraly立交桥视频;
“White Crow”音频;
YOLO目标检测器的局限和缺点
YOLO目标检测器的最大限制和缺点是:
它并不总能很好地处理小物体;
它尤其不适合处理密集的对象;
限制的原因是由于YOLO算法其本身:
YOLO对象检测器将输入图像划分为SxS网格,其中网格中的每个单元格仅预测单个对象;
如果单个单元格中存在多个小对象,则YOLO将无法检测到它们,最终导致错过对象检测;
因此,如果你的数据集是由许多靠近在一起的小对象组成时,那么就不应该使用YOLO算法。就小物体而言,更快的R-CNN往往效果最好,但是其速度也最慢。在这里也可以使用SSD算法, SSD通常在速度和准确性方面也有很好的权衡。
值得注意的是,在本教程中,YOLO比SSD运行速度慢,大约慢一个数量级。因此,如果你正在使用预先训练的深度学习对象检测器供OpenCV使用,可能需要考虑使用SSD算法而不是YOLO算法。
因此,在针对给定问题选择对象检测器时,我倾向于使用以下准则:
如果知道需要检测的是小物体并且速度方面不作求,我倾向于使用faster R-CNN算法;
如果速度是最重要的,我倾向于使用YOLO算法;
如果需要一个平衡的表现,我倾向于使用SSD算法;
想要训练自己的深度学习目标检测器?
在本教程中,使用的YOLO模型是在COCO数据集上预先训练的.。但是,如果想在自己的数据集上训练深度学习对象检测器,该如何操作呢?
大体思路是自己标注数据集,按照darknet网站上的指示及网上博客自己更改相应的参数训练即可。或者在我的书“ 深度学习计算机视觉与Python”中,详细讲述了如何将faster R-CNN、SSD和RetinaNet应用于:
检测图像中的徽标;
检测交通标志;
检测车辆的前视图和后视图(用于构建自动驾驶汽车应用);
检测图像和视频流中武器;
书中的所有目标检测章节都包含对算法和代码的详细说明,确保你能够成功训练自己的对象检测器。在这里可以了解有关我的书的更多信息(并获取免费的示例章节和目录)。
总结
在本教程中,我们学习了如何使用Deep Learning、OpenCV和Python完成YOLO对象检测。然后,我们简要讨论了YOLO架构,并用Python实现:
将YOLO对象检测应用于单个图像;
将YOLO对象检测应用于视频流;
在配备的3GHz Intel Xeon W处理器的机器上,YOLO的单次前向传输耗时约0.3秒; 但是,使用单次检测器(SSD),检测耗时只需0.03秒,速度提升了一个数量级。对于使用OpenCV和Python在CPU上进行基于实时深度学习的对象检测,你可能需要考虑使用SSD算法。