前言

YOLOV5模型从发布到现在都是炙手可热的目标检测模型，被广泛运用于各大场景之中。因此，我们不光要知道如何进行yolov5模型的训练，而且还要知道怎么进行部署应用。在本篇博客中，我将利用yolov5模型简单的实现从摄像头端到web端的部署应用demo，为读者提供一些部署思路。

一、YOLOV5的强大之处

你与目标检测高手之差一个YOLOV5模型。YOLOV5可以说是现目前几乎将所有目标检测tricks运用于一身的模型了。在它身上能找到很多目前主流的数据增强、模型训练、模型后处理的方法，下面我们就简单总结一下yolov5所使用到的方法：

• yolov5增加的功能：
  

• yolov5训练和预测的tricks：
  

二、YOLOV5部署多路摄像头的web应用

1.多路摄像头读取

在此篇博客中，采用了yolov5源码的datasets.py代码中的LoadStreams类进行多路摄像头视频流的读取。因为，我们只会用到datasets.py中视频流读取的部分代码，所以，将其提取出来，新建一个camera.py文件，下面则是camera.py文件的代码部分：

# coding:utf-8
import os
import cv2
import glob
import time
import numpy as np
from pathlib import Path
from utils.datasets import letterbox
from threading import Thread
from utils.general import clean_str


img_formats = ['bmp', 'jpg', 'jpeg', 'png', 'tif', 'tiff', 'dng', 'webp']  # acceptable image suffixes
vid_formats = ['mov', 'avi', 'mp4', 'mpg', 'mpeg', 'm4v', 'wmv', 'mkv']  # acceptable video suffixes


class LoadImages:  # for inference
    def __init__(self, path, img_size=640, stride=32):
        p = str(Path(path).absolute())  # os-agnostic absolute path
        if '*' in p:
            files = sorted(glob.glob(p, recursive=True))  # glob
        elif os.path.isdir(p):
            files = sorted(glob.glob(os.path.join(p, '*.*')))  # dir
        elif os.path.isfile(p):
            files = [p]  # files
        else:
            raise Exception(f'ERROR: {p} does not exist')

        images = [x for x in files if x.split('.')[-1].lower() in img_formats]
        videos = [x for x in files if x.split('.')[-1].lower() in vid_formats]
        ni, nv = len(images), len(videos)

        self.img_size = img_size
        self.stride = stride
        self.files = images + videos
        self.nf = ni + nv  # number of files
        self.video_flag = [False] * ni + [True] * nv
        self.mode = 'image'
        if any(videos):
            self.new_video(videos[0])  # new video
        else:
            self.cap = None
        assert self.nf > 0, f'No images or videos found in {p}. ' \
                            f'Supported formats are:\nimages: {img_formats}\nvideos: {vid_formats}'

    def __iter__(self):
        self.count = 0
        return self

    def __next__(self):
        if self.count == self.nf:
            raise StopIteration
        path = self.files[self.count]

        if self.video_flag[self.count]:
            # Read video
            self.mode = 'video'
            ret_val, img0 = self.cap.read()
            if not ret_val:
                self.count += 1
                self.cap.release()
                if self.count == self.nf:  # last video
                    raise StopIteration
                else:
                    path = self.files[self.count]
                    self.new_video(path)
                    ret_val, img0 = self.cap.read()

            self.frame += 1
            print(f'video {self.count + 1}/{self.nf} ({self.frame}/{self.nframes}) {path}: ', end='')

        else:
            # Read image
            self.count += 1
            img0 = cv2.imread(path)  # BGR
            assert img0 is not None, 'Image Not Found ' + path
            print(f'image {self.count}/{self.nf} {path}: ', end='')

        # Padded resize
        img = letterbox(img0, self.img_size, stride=self.stride)[0]

        # Convert
        img = img[:, :, ::-1].transpose(2, 0, 1)  # BGR to RGB, to 3x416x416
        img = np.ascontiguousarray(img)

        return path, img, img0, self.cap

    def new_video(self, path):
        self.frame = 0
        self.cap = cv2.VideoCapture(path)
        self.nframes = int(self.cap.get(cv2.CAP_PROP_FRAME_COUNT))

    def __len__(self):
        return self.nf  # number of files


class LoadWebcam:  # for inference
    def __init__(self, pipe='0', img_size=640, stride=32):
        self.img_size = img_size
        self.stride = stride

        if pipe.isnumeric():
            pipe = eval(pipe)  # local camera
        # pipe = 'rtsp://192.168.1.64/1'  # IP camera
        # pipe = 'rtsp://username:password@192.168.1.64/1'  # IP camera with login
        # pipe = 'http://wmccpinetop.axiscam.net/mjpg/video.mjpg'  # IP golf camera

        self.pipe = pipe
        self.cap = cv2.VideoCapture(pipe)  # video capture object
        self.cap.set(cv2.CAP_PROP_BUFFERSIZE, 3)  # set buffer size

    def __iter__(self):
        self.count = -1
        return self

    def __next__(self):
        self.count += 1
        if cv2.waitKey(1) == ord('q'):  # q to quit
            self.cap.release()
            cv2.destroyAllWindows()
            raise StopIteration

        # Read frame
        if self.pipe == 0:  # local camera
            ret_val, img0 = self.cap.read()
            img0 = cv2.flip(img0, 1)  # flip left-right
        else:  # IP camera
            n = 0
            while True:
                n += 1
                self.cap.grab()
                if n % 30 == 0:  # skip frames
                    ret_val, img0 = self.cap.retrieve()
                    if ret_val:
                        break

        # Print
        assert ret_val, f'Camera Error {self.pipe}'
        img_path = 'webcam.jpg'
        print(f'webcam {self.count}: ', end='')

        # Padded resize
        img = letterbox(img0, self.img_size, stride=self.stride)[0]

        # Convert
        img = img[:, :, ::-1].transpose(2, 0, 1)  # BGR to RGB, to 3x416x416
        img = np.ascontiguousarray(img)

        return img_path, img, img0, None

    def __len__(self):
        return 0


class LoadStreams:  # multiple IP or RTSP cameras
    def __init__(self, sources='streams.txt', img_size=640, stride=32):
        self.mode = 'stream'
        self.img_size = img_size
        self.stride = stride

        if os.path.isfile(sources):
            with open(sources, 'r') as f:
                sources = [x.strip() for x in f.read().strip().splitlines() if len(x.strip())]
        else:
            sources = [sources]

        n = len(sources)
        self.imgs = [None] * n
        self.sources = [clean_str(x) for x in sources]  # clean source names for later
        for i, s in enumerate(sources):
            # Start the thread to read frames from the video stream
            print(f'{i + 1}/{n}: {s}... ', end='')
            cap = cv2.VideoCapture(eval(s) if s.isnumeric() else s)
            assert cap.isOpened(), f'Failed to open {s}'
            w = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
            h = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
            fps = cap.get(cv2.CAP_PROP_FPS) % 100
            _, self.imgs[i] = cap.read()  # guarantee first frame
            thread = Thread(target=self.update, args=([i, cap]), daemon=True)
            print(f' success ({w}x{h} at {fps:.2f} FPS).')
            thread.start()
        print('')  # newline

        # check for common shapes
        s = np.stack([letterbox(x, self.img_size, stride=self.stride)[0].shape for x in self.imgs], 0)  # shapes
        self.rect = np.unique(s, axis=0).shape[0] == 1  # rect inference if all shapes equal
        if not self.rect:
            print('WARNING: Different stream shapes detected. For optimal performance supply similarly-shaped streams.')

    def update(self, index, cap):
        # Read next stream frame in a daemon thread
        n = 0
        while cap.isOpened():
            n += 1
            # _, self.imgs[index] = cap.read()
            cap.grab()
            if n == 4:  # read every 4th frame
                success, im = cap.retrieve()
                self.imgs[index] = im if success else self.imgs[index] * 0
                n = 0
            time.sleep(0.01)  # wait time

    def __iter__(self):
        self.count = -1
        return self

    def __next__(self):
        self.count += 1
        img0 = self.imgs.copy()
        if cv2.waitKey(1) == ord('q'):  # q to quit
            cv2.destroyAllWindows()
            raise StopIteration

        # Letterbox
        img = [letterbox(x, self.img_size, auto=self.rect, stride=self.stride)[0] for x in img0]

        # Stack
        img = np.stack(img, 0)

        # Convert
        img = img[:, :, :, ::-1].transpose(0, 3, 1, 2)  # BGR to RGB, to bsx3x416x416
        img = np.ascontiguousarray(img)

        return self.sources, img, img0, None

    def __len__(self):
        return 0  # 1E12 frames = 32 streams at 30 FPS for 30 years

2.模型封装

接下来，我们借助detect.py文件对yolov5模型进行接口封装，使其提供模型推理能力。新建一个yolov5.py文件，构建一个名为darknet的类，使用函数detect，提供目标检测能力。其代码如下：

# coding:utf-8
import cv2
import json
import time
import torch
import numpy as np
from camera import LoadStreams, LoadImages
from utils.torch_utils import select_device
from models.experimental import attempt_load
from utils.general import non_max_suppression, scale_coords, letterbox, check_imshow


class Darknet(object):
    """docstring for Darknet"""
    def __init__(self, opt):
        self.opt = opt
        self.device = select_device(self.opt["device"])
        self.half = self.device.type != 'cpu'  # half precision only supported on CUDA
        self.model = attempt_load(self.opt["weights"], map_location=self.device)
        self.stride = int(self.model.stride.max()) 
        self.model.to(self.device).eval()
        self.names = self.model.module.names if hasattr(self.model, 'module') else self.model.names
        if self.half: self.model.half()
        self.source = self.opt["source"]
        self.webcam = self.source.isnumeric() or self.source.endswith('.txt') or self.source.lower().startswith(
        ('rtsp://', 'rtmp://', 'http://'))
    
    def preprocess(self, img):
        img = np.ascontiguousarray(img)
        img = torch.from_numpy(img).to(self.device)
        img = img.half() if self.half else img.float()  # uint8 to fp16/32
        img /= 255.0  # 图像归一化
        if img.ndimension() == 3:
            img = img.unsqueeze(0)
        return img
    
    def detect(self, dataset):
        view_img = check_imshow()
        t0 = time.time()
        for path, img, img0s, vid_cap in dataset:
            img = self.preprocess(img)

            t1 = time.time()
            pred = self.model(img, augment=self.opt["augment"])[0]  # 0.22s
            pred = pred.float()
            pred = non_max_suppression(pred, self.opt["conf_thres"], self.opt["iou_thres"])
            t2 = time.time()

            pred_boxes = []
            for i, det in enumerate(pred):
                if self.webcam:  # batch_size >= 1
                    p, s, im0, frame = path[i], '%g: ' % i, img0s[i].copy(), dataset.count
                else:
                    p, s, im0, frame = path, '', img0s, getattr(dataset, 'frame', 0)
                s += '%gx%g ' % img.shape[2:]  # print string
                gn = torch.tensor(im0.shape)[[1, 0, 1, 0]]  # normalization gain whwh
                if det is not None and len(det):
                    det[:, :4] = scale_coords(
                        img.shape[2:], det[:, :4], im0.shape).round()

                    # Print results
                    for c in det[:, -1].unique():
                        n = (det[:, -1] == c).sum()  # detections per class
                        s += f"{n} {self.names[int(c)]}{'s' * (n > 1)}, "  # add to string

                    for *xyxy, conf, cls_id in det:
                        lbl = self.names[int(cls_id)]
                        xyxy = torch.tensor(xyxy).view(1, 4).view(-1).tolist()
                        score = round(conf.tolist(), 3)
                        label = "{}: {}".format(lbl, score)
                        x1, y1, x2, y2 = int(xyxy[0]), int(xyxy[1]), int(xyxy[2]), int(xyxy[3])
                        pred_boxes.append((x1, y1, x2, y2, lbl, score))
                        if view_img:
                            self.plot_one_box(xyxy, im0, color=(255, 0, 0), label=label)
                        
                # Print time (inference + NMS)
                # print(pred_boxes)
                print(f'{s}Done. ({t2 - t1:.3f}s)')

                if view_img:
                    print(str(p))
                    cv2.imshow(str(p), cv2.resize(im0, (800, 600)))
                    if self.webcam:
                        if cv2.waitKey(1) & 0xFF == ord('q'): break
                    else:
                    	cv2.waitKey(0)

        print(f'Done. ({time.time() - t0:.3f}s)')
        # print('[INFO] Inference time: {:.2f}s'.format(t3-t2))
        # return pred_boxes

    # Plotting functions
    def plot_one_box(self, x, img, color=None, label=None, line_thickness=None):
        # Plots one bounding box on image img
        tl = line_thickness or round(0.001 * max(img.shape[0:2])) + 1  # line thickness
        color = color or [random.randint(0, 255) for _ in range(3)]
        c1, c2 = (int(x[0]), int(x[1])), (int(x[2]), int(x[3]))
        cv2.rectangle(img, c1, c2, color, thickness=tl)
        if label:
            tf = max(tl - 1, 1)  # font thickness
            t_size = cv2.getTextSize(label, 0, fontScale=tl / 3, thickness=tf)[0]
            c2 = c1[0] + t_size[0], c1[1] - t_size[1] - 3
            cv2.rectangle(img, c1, c2, color, -1)  # filled
            cv2.putText(img, label, (c1[0], c1[1] - 2), 0, tl / 3, [0, 0, 0], thickness=tf, lineType=cv2.LINE_AA)


if __name__ == "__main__":
    with open('yolov5_config.json', 'r', encoding='utf8') as fp:
        opt = json.load(fp)
        print('[INFO] YOLOv5 Config:', opt)
    darknet = Darknet(opt)
    if darknet.webcam:
        # cudnn.benchmark = True  # set True to speed up constant image size inference
        dataset = LoadStreams(darknet.source, img_size=opt["imgsz"], stride=darknet.stride)
    else:
        dataset = LoadImages(darknet.source, img_size=opt["imgsz"], stride=darknet.stride)
    darknet.detect(dataset)
    cv2.destroyAllWindows()

此外，还需要提供一个模型配置文件，我们使用json文件进行保存。新建一个yolov5_config.json文件，内容如下：

{
	"source": "streams.txt",  # 为视频图像文件地址
	"weights": "runs/train/exp/weights/best.pt", # 自己的模型地址
	"device": "cpu", # 使用的device类别，如是GPU，可填"0"
	"imgsz": 640,  # 输入图像的大小
	"stride": 32,  # 步长
	"conf_thres": 0.35, # 置信值阈值
	"iou_thres": 0.45,  # iou阈值
	"augment": false  # 是否使用图像增强
}

视频图像文件可以是单独的一张图像，如:"…/images/demo.jpg"，也可以是一个视频文件，如："…/videos/demo.mp4"，也可以是一个视频流地址，如：“rtsp://wowzaec2demo.streamlock.net/vod/mp4:BigBuckBunny_115k.mov”，还可以是一个txt文件，里面包含多个视频流地址，如：

rtsp://wowzaec2demo.streamlock.net/vod/mp4:BigBuckBunny_115k.mov
rtsp://wowzaec2demo.streamlock.net/vod/mp4:BigBuckBunny_115k.mov

- 有了如此配置信息，通过运行yolov5.py代码，我们能实现对视频文件（mp4、avi等）、视频流地址（http、rtsp、rtmp等）、图片（jpg、png）等视频图像文件进行目标检测推理的效果。

3.Flask后端处理

有了对模型封装的代码，我们就可以利用flask框架实时向前端推送算法处理之后的图像了。新建一个web_main.py文件：

# import the necessary packages
from yolov5 import Darknet
from camera import LoadStreams, LoadImages
from utils.general import non_max_suppression, scale_coords, letterbox, check_imshow
from flask import Response
from flask import Flask
from flask import render_template
import time
import torch
import json
import cv2
import os



# initialize a flask object
app = Flask(__name__)

# initialize the video stream and allow the camera sensor to warmup
with open('yolov5_config.json', 'r', encoding='utf8') as fp:
    opt = json.load(fp)
    print('[INFO] YOLOv5 Config:', opt)

darknet = Darknet(opt)
if darknet.webcam:
    # cudnn.benchmark = True  # set True to speed up constant image size inference
    dataset = LoadStreams(darknet.source, img_size=opt["imgsz"], stride=darknet.stride)
else:
    dataset = LoadImages(darknet.source, img_size=opt["imgsz"], stride=darknet.stride)
time.sleep(2.0)

@app.route("/")
def index():
    # return the rendered template
    return render_template("index.html")

def detect_gen(dataset, feed_type):
    view_img = check_imshow()
    t0 = time.time()
    for path, img, img0s, vid_cap in dataset:
        img = darknet.preprocess(img)

        t1 = time.time()
        pred = darknet.model(img, augment=darknet.opt["augment"])[0]  # 0.22s
        pred = pred.float()
        pred = non_max_suppression(pred, darknet.opt["conf_thres"], darknet.opt["iou_thres"])
        t2 = time.time()

        pred_boxes = []
        for i, det in enumerate(pred):
            if darknet.webcam:  # batch_size >= 1
                feed_type_curr, p, s, im0, frame = "Camera_%s" % str(i), path[i], '%g: ' % i, img0s[i].copy(), dataset.count
            else:
                feed_type_curr, p, s, im0, frame = "Camera", path, '', img0s, getattr(dataset, 'frame', 0)

            s += '%gx%g ' % img.shape[2:]  # print string
            gn = torch.tensor(im0.shape)[[1, 0, 1, 0]]  # normalization gain whwh
            if det is not None and len(det):
                det[:, :4] = scale_coords(
                    img.shape[2:], det[:, :4], im0.shape).round()

                # Print results
                for c in det[:, -1].unique():
                    n = (det[:, -1] == c).sum()  # detections per class
                    s += f"{n} {darknet.names[int(c)]}{'s' * (n > 1)}, "  # add to string

                for *xyxy, conf, cls_id in det:
                    lbl = darknet.names[int(cls_id)]
                    xyxy = torch.tensor(xyxy).view(1, 4).view(-1).tolist()
                    score = round(conf.tolist(), 3)
                    label = "{}: {}".format(lbl, score)
                    x1, y1, x2, y2 = int(xyxy[0]), int(xyxy[1]), int(xyxy[2]), int(xyxy[3])
                    pred_boxes.append((x1, y1, x2, y2, lbl, score))
                    if view_img:
                        darknet.plot_one_box(xyxy, im0, color=(255, 0, 0), label=label)

            # Print time (inference + NMS)
            # print(pred_boxes)
            print(f'{s}Done. ({t2 - t1:.3f}s)')
            if feed_type_curr == feed_type:
                frame = cv2.imencode('.jpg', im0)[1].tobytes()
                yield (b'--frame\r\n' b'Content-Type: image/jpeg\r\n\r\n' + frame + b'\r\n')

@app.route('/video_feed/<feed_type>')
def video_feed(feed_type):
    """Video streaming route. Put this in the src attribute of an img tag."""
    if feed_type == 'Camera_0':
        return Response(detect_gen(dataset=dataset, feed_type=feed_type),
                        mimetype='multipart/x-mixed-replace; boundary=frame')

    elif feed_type == 'Camera_1':
        return Response(detect_gen(dataset=dataset, feed_type=feed_type),
                        mimetype='multipart/x-mixed-replace; boundary=frame')


if __name__ == '__main__':
    app.run(host='0.0.0.0', port="5000", threaded=True)

通过detect_gen函数将多个视频流地址推理后的图像按照feed_type类型，通过video_feed视频流路由进行传送到前端。

4.前端展示

最后，我们写一个简单的前端代码。首先新建一个templates文件夹，再在此文件夹中新建一个index.html文件，将下面h5代码写入其中：

<html>
  <head>
	<style>
	* {
	  box-sizing: border-box;
	  text-align: center;
	}

	.img-container {
	  float: left;
	  width: 30%;
	  padding: 5px;
	}

	.clearfix::after {
	  content: "";
	  clear: both;
	  display: table;
	}
	.clearfix{
		margin-left: 500px;
	}
	</style>
  </head>
  <body>
  	<h1>Multi-camera with YOLOv5</h1>
  	<div class="clearfix">
  		<div class="img-container" align="center">
        	<p align="center">Live stream 1</p>
        	<img src="{{ url_for('video_feed', feed_type='Camera_0') }}" class="center"  style="border:1px solid black;width:100%" alt="Live Stream 1">
  		</div>
  		<div class="img-container" align="center">
        	<p align="center">Live stream 2</p>
        	<img src="{{ url_for('video_feed', feed_type='Camera_1') }}" class="center"  style="border:1px solid black;width:100%" alt="Live Stream 2">
  		</div>
	</div>
  </body>
</html>

至此，我们利用YOLOv5模型实现多路摄像头实时推理代码就写完了，下面我们开始运行：

- 在终端中进行跟目录下，直接运行：

python web_main.py

然后，会在终端中出现如下信息：

[INFO] YOLOv5 Config: {'source': 'streams.txt', 'weights': 'runs/train/exp/weights/best.pt', 'device': 'cpu', 'imgsz': 640, 'stride': 32, 'conf_thres': 0.35, 'iou_thres': 0.45, 'augment': False}
Fusing layers...
1/2: rtsp://wowzaec2demo.streamlock.net/vod/mp4:BigBuckBunny_115k.mov...  success (240x160 at 24.00 FPS).
2/2: rtsp://wowzaec2demo.streamlock.net/vod/mp4:BigBuckBunny_115k.mov...  success (240x160 at 24.00 FPS).

 * Serving Flask app "web_main" (lazy loading)
 * Environment: production
   WARNING: This is a development server. Do not use it in a production deployment.
   Use a production WSGI server instead.
 * Debug mode: off
 * Running on http://0.0.0.0:5000/ (Press CTRL+C to quit)

* 接着打开浏览器，输入localhost:5000后，终端没有报任何错误，则就会出现如下页面：

总结

1. 由于没有额外的视频流rtmp/rtsp文件地址，所以就找了一个公开的视频流地址，但是没有办法看到检测效果；
2. 部署的时候，只能使用视频流地址进行推理，且可以为多个视频流地址，保存为stream.txt，用yolov5_config.json导入；
3. 此demo版本为简易版的端到端模型部署方案，还可以根据场景需要添加更多功能。

版权声明：本文为CSDN博主「llh_1178」的原创文章，遵循CC 4.0 BY-SA版权协议，转载请附上原文出处链接及本声明。
原文链接：https://blog.csdn.net/llh_1178/article/details/115075941