Yolov5训练自己的模型（检测人和安全帽）

一. 准备

• 下载VOC数据或者自己收集的图片
• VOC官网
• 确保有jpg和xml文件
• 生成train.txt，val.txt，test.txt和trainval.txt四个文件，存放训练集、验证集、测试集图片的名字（无后缀.jpg）

        参考代码：split_train_val.py

# coding:utf-8

import os
import random
import argparse

parser = argparse.ArgumentParser()
#xml文件的地址，根据自己的数据进行修改 xml一般存放在Annotations下
parser.add_argument('--xml_path', default='Annotations', type=str, help='input xml label path')
#数据集的划分，地址选择自己数据下的ImageSets/Main
parser.add_argument('--txt_path', default='ImageSets/Main', type=str, help='output txt label path')
opt = parser.parse_args()

trainval_percent = 1.0
train_percent = 0.9
xmlfilepath = opt.xml_path
txtsavepath = opt.txt_path
total_xml = os.listdir(xmlfilepath)
if not os.path.exists(txtsavepath):
    os.makedirs(txtsavepath)

num = len(total_xml)
list_index = range(num)
tv = int(num * trainval_percent)
tr = int(tv * train_percent)
trainval = random.sample(list_index, tv)
train = random.sample(trainval, tr)

file_trainval = open(txtsavepath + '/trainval.txt', 'w')
file_test = open(txtsavepath + '/test.txt', 'w')
file_train = open(txtsavepath + '/train.txt', 'w')
file_val = open(txtsavepath + '/val.txt', 'w')

for i in list_index:
    name = total_xml[i][:-4] + '\n'
    if i in trainval:
        file_trainval.write(name)
        if i in train:
            file_train.write(name)
        else:
            file_val.write(name)
    else:
        file_test.write(name)

file_trainval.close()
file_train.close()
file_val.close()
file_test.close()

• 将labels（xml文件）转换为yolo_txt文件
  • 文件每一行为一个目标的信息，包括class, x_center, y_center, width, height格式

        参考代码：voc_label.py

# -*- coding: utf-8 -*-
import xml.etree.ElementTree as ET
import os
from os import getcwd

sets = ['train', 'val', 'test']
classes = ["hat","person"]   # 改成自己的类别
abs_path = os.getcwd()
print(abs_path)

def convert(size, box):
    dw = 1. / (size[0])
    dh = 1. / (size[1])
    x = (box[0] + box[1]) / 2.0 - 1
    y = (box[2] + box[3]) / 2.0 - 1
    w = box[1] - box[0]
    h = box[3] - box[2]
    x = x * dw
    w = w * dw
    y = y * dh
    h = h * dh
    return x, y, w, h

def convert_annotation(image_id):
    in_file = open('/Users/cosmomu/Desktop/检测/yolov5/VOC2028/Annotations/%s.xml' % (image_id), encoding='UTF-8')
    out_file = open('/Users/cosmomu/Desktop/检测/yolov5/VOC2028/labels/%s.txt' % (image_id), 'w')
    tree = ET.parse(in_file)
    root = tree.getroot()
    size = root.find('size')
    w = int(size.find('width').text)
    h = int(size.find('height').text)
    for obj in root.iter('object'):
        # difficult = obj.find('difficult').text
        # difficult = obj.find('Difficult').text
        cls = obj.find('name').text
        # if cls not in classes or int(difficult) == 1:
        #     continue
        cls_id = classes.index(cls)
        xmlbox = obj.find('bndbox')
        b = (float(xmlbox.find('xmin').text), float(xmlbox.find('xmax').text), float(xmlbox.find('ymin').text),
             float(xmlbox.find('ymax').text))
        b1, b2, b3, b4 = b
        # 标注越界修正
        if b2 > w:
            b2 = w
        if b4 > h:
            b4 = h
        b = (b1, b2, b3, b4)
        bb = convert((w, h), b)
        out_file.write(str(cls_id) + " " + " ".join([str(a) for a in bb]) + '\n')

wd = getcwd()
for image_set in sets:
    if not os.path.exists('/Users/cosmomu/Desktop/检测/yolov5/VOC2028/labels/'):
        os.makedirs('/Users/cosmomu/Desktop/检测/yolov5/VOC2028/labels/')
    image_ids = open('/Users/cosmomu/Desktop/检测/yolov5/VOC2028/ImageSets/Main/%s.txt' % (image_set)).read().strip().split()
    list_file = open('/Users/cosmomu/Desktop/检测/yolov5/VOC2028/%s.txt' % (image_set), 'w')
    for image_id in image_ids:
        list_file.write('/Users/cosmomu/Desktop/检测/yolov5/VOC2028/JPEGImages/%s.jpg\n' % (image_id))
        convert_annotation(image_id)
    list_file.close()

运行voc_label.py时报错“ZeroDivisionError: float division by zero”的原因是：标注文件中存在width为0或者height为0的数据。

• 在yolov5目录下的data文件夹下新建一个你要训练的类型.yaml文件，这里是helmet.yaml

参考代码：helmet.yaml

train: /Users/cosmomu/Desktop/检测/yolov5/VOC2028/
val: /Users/cosmomu/Desktop/检测/yolov5/VOC2028/

# number of classes
nc: 2

# class names
names: ['hat', 'person']

二. 训练

• 如果你没有GPU，可以在最前面加上这句话     os.environ["CUDA_VISIBLE_DEVICES"] = "-1"  即采用CPU进行训练
• 聚类得出先验框（可选步骤）

参考代码：kmeas.py

import numpy as np

def iou(box, clusters):
    """
    Calculates the Intersection over Union (IoU) between a box and k clusters.
    :param box: tuple or array, shifted to the origin (i. e. width and height)
    :param clusters: numpy array of shape (k, 2) where k is the number of clusters
    :return: numpy array of shape (k, 0) where k is the number of clusters
    """
    x = np.minimum(clusters[:, 0], box[0])
    y = np.minimum(clusters[:, 1], box[1])
    if np.count_nonzero(x == 0) > 0 or np.count_nonzero(y == 0) > 0:
        raise ValueError("Box has no area")                 # 如果报这个错，可以把这行改成pass即可

    intersection = x * y
    box_area = box[0] * box[1]
    cluster_area = clusters[:, 0] * clusters[:, 1]

    iou_ = intersection / (box_area + cluster_area - intersection)

    return iou_

def avg_iou(boxes, clusters):
    """
    Calculates the average Intersection over Union (IoU) between a numpy array of boxes and k clusters.
    :param boxes: numpy array of shape (r, 2), where r is the number of rows
    :param clusters: numpy array of shape (k, 2) where k is the number of clusters
    :return: average IoU as a single float
    """
    return np.mean([np.max(iou(boxes[i], clusters)) for i in range(boxes.shape[0])])

def translate_boxes(boxes):
    """
    Translates all the boxes to the origin.
    :param boxes: numpy array of shape (r, 4)
    :return: numpy array of shape (r, 2)
    """
    new_boxes = boxes.copy()
    for row in range(new_boxes.shape[0]):
        new_boxes[row][2] = np.abs(new_boxes[row][2] - new_boxes[row][0])
        new_boxes[row][3] = np.abs(new_boxes[row][3] - new_boxes[row][1])
    return np.delete(new_boxes, [0, 1], axis=1)


def kmeans(boxes, k, dist=np.median):
    """
    Calculates k-means clustering with the Intersection over Union (IoU) metric.
    :param boxes: numpy array of shape (r, 2), where r is the number of rows
    :param k: number of clusters
    :param dist: distance function
    :return: numpy array of shape (k, 2)
    """
    rows = boxes.shape[0]

    distances = np.empty((rows, k))
    last_clusters = np.zeros((rows,))

    np.random.seed()

    # the Forgy method will fail if the whole array contains the same rows
    clusters = boxes[np.random.choice(rows, k, replace=False)]

    while True:
        for row in range(rows):
            distances[row] = 1 - iou(boxes[row], clusters)

        nearest_clusters = np.argmin(distances, axis=1)

        if (last_clusters == nearest_clusters).all():
            break

        for cluster in range(k):
            clusters[cluster] = dist(boxes[nearest_clusters == cluster], axis=0)

        last_clusters = nearest_clusters

    return clusters

if __name__ == '__main__':
    a = np.array([[1, 2, 3, 4], [5, 7, 6, 8]])
    print(translate_boxes(a))

参考代码：clauculate_anchors.py 

# -*- coding: utf-8 -*-
# 根据标签文件求先验框

import os
import numpy as np
import xml.etree.cElementTree as et
from kmeans import kmeans, avg_iou

FILE_ROOT = "yolov5/VOC2028/"     # 根路径
ANNOTATION_ROOT = "Annotations"  # 数据集标签文件夹路径
ANNOTATION_PATH = FILE_ROOT + ANNOTATION_ROOT

ANCHORS_TXT_PATH = "yolov5/data/anchors.txt"

CLUSTERS = 9
CLASS_NAMES = ['hat', 'person']

def load_data(anno_dir, class_names):
    xml_names = os.listdir(anno_dir)
    boxes = []
    for xml_name in xml_names:
        xml_pth = os.path.join(anno_dir, xml_name)
        tree = et.parse(xml_pth)

        width = float(tree.findtext("./size/width"))
        height = float(tree.findtext("./size/height"))

        for obj in tree.findall("./object"):
            cls_name = obj.findtext("name")
            if cls_name in class_names:
                xmin = float(obj.findtext("bndbox/xmin")) / width
                ymin = float(obj.findtext("bndbox/ymin")) / height
                xmax = float(obj.findtext("bndbox/xmax")) / width
                ymax = float(obj.findtext("bndbox/ymax")) / height

                box = [xmax - xmin, ymax - ymin]
                boxes.append(box)
            else:
                continue
    return np.array(boxes)

if __name__ == '__main__':

    anchors_txt = open(ANCHORS_TXT_PATH, "w")

    train_boxes = load_data(ANNOTATION_PATH, CLASS_NAMES)
    count = 1
    best_accuracy = 0
    best_anchors = []
    best_ratios = []

    for i in range(10):      ##### 可以修改，不要太大，否则时间很长
        anchors_tmp = []
        clusters = kmeans(train_boxes, k=CLUSTERS)
        idx = clusters[:, 0].argsort()
        clusters = clusters[idx]
        # print(clusters)

        for j in range(CLUSTERS):
            anchor = [round(clusters[j][0] * 640, 2), round(clusters[j][1] * 640, 2)]
            anchors_tmp.append(anchor)
            print(f"Anchors:{anchor}")

        temp_accuracy = avg_iou(train_boxes, clusters) * 100
        print("Train_Accuracy:{:.2f}%".format(temp_accuracy))

        ratios = np.around(clusters[:, 0] / clusters[:, 1], decimals=2).tolist()
        ratios.sort()
        print("Ratios:{}".format(ratios))
        print(20 * "*" + " {} ".format(count) + 20 * "*")

        count += 1

        if temp_accuracy > best_accuracy:
            best_accuracy = temp_accuracy
            best_anchors = anchors_tmp
            best_ratios = ratios

    anchors_txt.write("Best Accuracy = " + str(round(best_accuracy, 2)) + '%' + "\r\n")
    anchors_txt.write("Best Anchors = " + str(best_anchors) + "\r\n")
    anchors_txt.write("Best Ratios = " + str(best_ratios))
    anchors_txt.close()

根据clauculate_anchors.py 生成新anchors文件在指定目录下

（如果自己动手生成anchors时候报错“ raise ValueError(“Box has no area”)”，可能是标注数据中有的目标物很小很小，可以把kmeans.py中第13行注释掉，改成pass即可解决这个报错。）

运行结果如下：

Best Accuracy = 80.97%
Best Anchors = [[7.8, 15.96], [9.52, 20.47], [11.86, 24.41], [15.13, 30.02], [20.5, 38.0], [29.44, 50.53], [43.0, 72.82], [67.41, 115.2], [125.87, 215.49]]
Best Ratios = [0.46, 0.49, 0.49, 0.5, 0.54, 0.58, 0.58, 0.59, 0.59]

• 根据目标需求选择需要的结构（s, m, l, x, n, s6, m6, l6, x6, n6），这里选择5s或5m，先用5s测试，因为结构大小（也意味着训练时间长短随着nsmlx增大）。如果进行了上一步，需要对anchors.txt文件内数值进行取整。
• 将选取的结构yaml文件中类别数改成对应数量。
• (如果生成了新的anchors文件，则更新anchors数据）

参考代码：yolov5s.yaml

# YOLOv5  by Ultralytics, GPL-3.0 license

# Parameters
nc: 2  # number of classes
depth_multiple: 0.33  # model depth multiple
width_multiple: 0.50  # layer channel multiple
anchors:
  - [10,13, 16,30, 33,23]  # P3/8
  - [30,61, 62,45, 59,119]  # P4/16
  - [116,90, 156,198, 373,326]  # P5/32

# YOLOv5 v6.0 backbone
backbone:
  # [from, number, module, args]
  [[-1, 1, Conv, [64, 6, 2, 2]],  # 0-P1/2
   [-1, 1, Conv, [128, 3, 2]],  # 1-P2/4
   [-1, 3, C3, [128]],
   [-1, 1, Conv, [256, 3, 2]],  # 3-P3/8
   [-1, 6, C3, [256]],
   [-1, 1, Conv, [512, 3, 2]],  # 5-P4/16
   [-1, 9, C3, [512]],
   [-1, 1, Conv, [1024, 3, 2]],  # 7-P5/32
   [-1, 3, C3, [1024]],
   [-1, 1, SPPF, [1024, 5]],  # 9
  ]

# YOLOv5 v6.0 head
head:
  [[-1, 1, Conv, [512, 1, 1]],
   [-1, 1, nn.Upsample, [None, 2, 'nearest']],
   [[-1, 6], 1, Concat, [1]],  # cat backbone P4
   [-1, 3, C3, [512, False]],  # 13

   [-1, 1, Conv, [256, 1, 1]],
   [-1, 1, nn.Upsample, [None, 2, 'nearest']],
   [[-1, 4], 1, Concat, [1]],  # cat backbone P3
   [-1, 3, C3, [256, False]],  # 17 (P3/8-small)

   [-1, 1, Conv, [256, 3, 2]],
   [[-1, 14], 1, Concat, [1]],  # cat head P4
   [-1, 3, C3, [512, False]],  # 20 (P4/16-medium)

   [-1, 1, Conv, [512, 3, 2]],
   [[-1, 10], 1, Concat, [1]],  # cat head P5
   [-1, 3, C3, [1024, False]],  # 23 (P5/32-large)

   [[17, 20, 23], 1, Detect, [nc, anchors]],  # Detect(P3, P4, P5)
  ]

• 更改train.py的default

参考代码：train.py

def parse_opt(known=False):
    parser = argparse.ArgumentParser()
    parser.add_argument('--weights', type=str, default=ROOT / 'yolov5s.pt', help='initial weights path')
    parser.add_argument('--cfg', type=str, default='models/yolov5s/yaml', help='model.yaml path')
    parser.add_argument('--data', type=str, default=ROOT / 'data/helmet.yaml', help='dataset.yaml path')
    parser.add_argument('--hyp', type=str, default=ROOT / 'data/hyps/hyp.scratch.yaml', help='hyperparameters path')
    parser.add_argument('--epochs', type=int, default=300)
    parser.add_argument('--batch-size', type=int, default=16, help='total batch size for all GPUs, -1 for autobatch')
    parser.add_argument('--imgsz', '--img', '--img-size', type=int, default=640, help='train, val image size (pixels)')
    parser.add_argument('--rect', action='store_true', help='rectangular training')
    parser.add_argument('--resume', nargs='?', const=True, default=False, help='resume most recent training')
    parser.add_argument('--nosave', action='store_true', help='only save final checkpoint')
    parser.add_argument('--noval', action='store_true', help='only validate final epoch')
    parser.add_argument('--noautoanchor', action='store_true', help='disable autoanchor check')
    parser.add_argument('--evolve', type=int, nargs='?', const=300, help='evolve hyperparameters for x generations')
    parser.add_argument('--bucket', type=str, default='', help='gsutil bucket')
    parser.add_argument('--cache', type=str, nargs='?', const='ram', help='--cache images in "ram" (default) or "disk"')
    parser.add_argument('--image-weights', action='store_true', help='use weighted image selection for training')
    parser.add_argument('--device', default='', help='cuda device, i.e. 0 or 0,1,2,3 or cpu')
    parser.add_argument('--multi-scale', action='store_true', help='vary img-size +/- 50%%')
    parser.add_argument('--single-cls', action='store_true', help='train multi-class data as single-class')
    parser.add_argument('--adam', action='store_true', help='use torch.optim.Adam() optimizer')
    parser.add_argument('--sync-bn', action='store_true', help='use SyncBatchNorm, only available in DDP mode')
    parser.add_argument('--workers', type=int, default=8, help='max dataloader workers (per RANK in DDP mode)')
    parser.add_argument('--project', default=ROOT / 'runs/train', help='save to project/name')
    parser.add_argument('--name', default='exp', help='save to project/name')
    parser.add_argument('--exist-ok', action='store_true', help='existing project/name ok, do not increment')
    parser.add_argument('--quad', action='store_true', help='quad dataloader')
    parser.add_argument('--linear-lr', action='store_true', help='linear LR')
    parser.add_argument('--label-smoothing', type=float, default=0.0, help='Label smoothing epsilon')
    parser.add_argument('--patience', type=int, default=100, help='EarlyStopping patience (epochs without improvement)')
    parser.add_argument('--freeze', nargs='+', type=int, default=[0], help='Freeze layers: backbone=10, first3=0 1 2')
    parser.add_argument('--save-period', type=int, default=-1, help='Save checkpoint every x epochs (disabled if < 1)')
    parser.add_argument('--local_rank', type=int, default=-1, help='DDP parameter, do not modify')

注意更改其中的weights, cfg, data, 参数含义如下：

epochs：指的就是训练过程中整个数据集将被迭代多少次,显卡不行你就调小点。
batch-size：一次看完多少张图片才进行权重更新，梯度下降的mini-batch,显卡不行你就调小点。
cfg：存储模型结构的配置文件
data：存储训练、测试数据的文件
img-size：输入图片宽高,显卡不行你就调小点。
rect：进行矩形训练
resume：恢复最近保存的模型开始训练
nosave：仅保存最终checkpoint
notest：仅测试最后的epoch
evolve：进化超参数
bucket：gsutil bucket
cache-images：缓存图像以加快训练速度
weights：权重文件路径
name： 重命名results.txt to results_name.txt
device：cuda device, i.e. 0 or 0,1,2,3 or cpu
adam：使用adam优化
multi-scale：多尺度训练，img-size +/- 50%
single-cls：单类别的训练集

• 训练命令：

  python train.py --img 640 --batch 16 --epoch 300 --data data/helmet.yaml --cfg models/yolov5s.yaml --weights weights/yolov5s.pt --device '0'
  

  如果Cuda版本不对（不是>=10.1版本），在调用GPU训练时会报错。去各种torch下载合适的版本并用conda安装。

• 查看当前版本

功能：查看机器上GPU情况

命令： nvidia-smi

功能：显示机器上gpu的情况

命令： nvidia-smi -l

功能：定时更新显示机器上gpu的情况

命令：watch -n 3 nvidia-smi

功能：设定刷新时间（秒）显示GPU使用情况

•  训练时可视化：pip install wandb

 

•  训练遇到的问题：AssertionError: train: No labels in/data/yolov5/VOC2028/JPEGImages.cache.

解决方法：

打开dataset.py文件，使用快捷键Ctrl+F使用搜索框搜索define label

按照正常的VOC标注之后图片应该时保存在JPEGImages文件夹下的，但是根据源码则是读取的images里的图片，因此需要将images改为JPEGImages，这样就能正常读取了。

• 训练结束以后会在runs/train/exp中生成一个weights文件夹里面有last.pt和best.pt。

三. 测试 

评估模型好坏就是在有标注的测试集或者验证集上进行模型效果的评估，在目标检测中最常使用的评估指标为mAP。在test.py文件中指定数据集配置文件和训练结果模型。

python detect.py --weights weights/best.pt --source test/test.mp4

随便找了一个视频做了个测试还不错。

正常应该用test.py来进行测试，有点懒了直接用了。

python test.py  --data data/helmet.yaml --weights weights/best.pt --augment

附上我学习的帖子：

YOLOv5训练自己的数据集（超详细完整版）

解决torch.cuda.is_available()返回结果为False

【AI实战】动手训练自己的目标检测模型（YOLO篇）

YOLOv5模型训练