一、YOLO算法简介
YOLO是一个基于卷积神经网络的实时目标检测算法，基于darknet框架实现，关于YOLO算法的理论知识可以参考YOLO的官方网站及paper，另外还可以通过吴恩达的深度学习课程Convolutional Neural Networks，学习YOLO算法的原理。本文的目的是通过作者提供的代码，实现自己的数据集的模型训练过程。
二、数据集准备
首先需要准备自己的数据集，本文以单一类型识别为例，介绍如何进行单一分类的数据训练。作者官网有介绍如何在VOC格式的数据集下进行训练，关于VOC数据集的格式可以参考这篇博客PASCAL VOC数据集分析,笔者尝试训练了VOC数据集的20种分类数据，在Ubuntu16.04LTS+2G显存GTX960m显卡上训练，迭代45000次大概用时3天4夜，得出了初步的效果，所以不建议初次进行训练的人使用VOC数据集进行训练。笔者本次训练使用的数据集为食物图片数据集，初步选取了300张图片。接下来便是对数据进行标注。
三、数据标记
YOLOv3目标检测需要对要识别的数据以最小包围矩阵进行标记，需要中心点及四个顶点的像素坐标作为图片的标签。此处推荐一个标注工具LabelImg 具体的安装使用方法在作者的github上已经写得很清楚，此处不再赘述。该工具标注后的标签是标准的VOC数据集的XML格式标签，文件名和图片名一致，这样方便后续我们的转换，我们要记住标签文件的保存路径，笔者以自己的为例，图片文件存放于/home/wingc/SourceCode/food_detection/darknet/data/obj。标签文件存放于 /home/wingc/SourceCode/food_detection/darknet/data/obj/annotations。标签文件的格式如下所示
<annotation>
<folder>Image</folder>
<filename>0000</filename>
<source>
<database>Pedestrian_ultrared</database>
</source>
<size>
<width>384</width>
<height>288</height>
</size>
<object>
<name>n00000001</name>
<bndbox>
<xmin>232</xmin>
<xmax>248</xmax>
<ymin>161</ymin>
<ymax>203</ymax>
</bndbox>
</object>
</annotation>
四、标签格式转换
第四步就是将标记好的xml格式的标签转化为txt格式，这里提供一个脚本进行格式的转化，具体的文件路径根据个人的不同配置进行修改。
import xml.etree.ElementTree as ET
import glob
import os
sets = [('2012', 'train'), ('2012', 'val'), ('2007', 'train'), ('2007', 'val'), ('2007', 'test')]
classes = ["Shredded cabbage"]
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
# 图片所在路径及图片名
def convert_annotation(base_path, img_id):
in_file = open(os.path.join(base_path, 'annotations/{}.xml'.format(img_id)), 'r')
out_file = open(os.path.join(base_path, 'labels/{}.txt'.format(img_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
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))
bb = convert((w, h), b)
out_file.write(str(cls_id) + " " + " ".join([str(a) for a in bb]) + '\n')
def generate_train_test_txt(work_path, xml_path, train_path, test_path, test_percentage):
train_file = open(train_path, 'w')
test_file = open(test_path, 'w')
counter = 1
test_index = round(100 / test_percentage)
for xml_file_path in glob.iglob(os.path.join(xml_path, "*.xml")):
title, ext = os.path.splitext(os.path.basename(xml_file_path))
if counter == test_index:
counter = 1
test_file.write(work_path + '/' + title + '.jpg' + '\n')
else:
train_file.write(work_path + '/' + title + '.jpg' + '\n')
counter += 1
train_file.close()
test_file.close()
if __name__ == '__main__':
wd = input("请输入图片所在文件夹\n")
xml_wd = os.path.join(wd, 'annotations')
label_wd = os.path.join(wd, 'labels')
train_txt = os.path.join(wd, 'train.txt')
test_txt = os.path.join(wd, 'test.txt')
# 划分训练集与测试集并写入文件
generate_train_test_txt(wd, xml_wd, train_txt, test_txt, 10)
# 将标签数据放入labels文件夹
if not os.path.exists(os.path.join(wd, 'labels')):
os.makedirs(os.path.join(wd, 'labels'))
train_img_paths = open(train_txt, 'r').read().strip().split()
test_img_paths = open(test_txt, 'r').read().strip().split()
img_paths = train_img_paths + test_img_paths
for img_path in img_paths:
convert_annotation(wd, os.path.splitext(os.path.basename(img_path))[0])
需要注意的是如果数据集格式不是按照PASCAL VOC数据集来组织的，那么txt格式的标签需要保存到和图片文件同一目录下,原因在于代码中会根据图片路径读取标签路径，代码在data.c的fill_truth_detection函数中，具体的代码分析可参考博客YOLOv2代码分析_读取labels{by zhangzexuan}
五、数据集的划分
接下来我们就要将数据集划分为训练集和测试集，训练集的图片路径存放在train.txt中，一行为一张图片的路径,测试集的图片路径存放在test.txt中，数据集划分的脚本在第四部分已经给出，训练集与测试集的比例在代码中可进行调整，同时由于笔者只进行了部分图片标签的标注，所以是根据已标记的XML格式标签的文件名进行划分。
六、配置文件的准备
当以上的准备工作完成后，最后一步就是进行配置文件的配置，这也是最重要的一步，直接决定你训练的成败与否。我们需要创建三个配置文件
- cfg/obj.data
- cfg/obj.names
- cfg/yolo-obj.cfg
obj.data文件指定了划分的种类，训练集与测试集的路径，要识别的物体的名字，及训练过程中权重文件的保存路径。
classes= 1
train = train.txt
valid = test.txt
names = obj.names
backup = backup/
obj.names中存放要标记的物体的名称，每行对应一种类别,这里注意到只指定了图片的路径而没有准备标签的路径,原因在第四节已经解释,标签路径会根据图片路径来寻找,例如:
图片路径为:/home/project/images/test.jpg或/home/project/JEPGimages/test.jpg或/home/project/任意名称/test.jpg
标签路径为:/home/project/labels/test.txt或/home/project/labels/test.txt或/home/project/任意名称/test.txt
Shredded Cabbage
yolo-obj.cfg最为重要，改配置文件定义了一些超参数，以下说几个重点的地方：
- batch与subdivisions要注意测试的时候和训练的时候要分别注释掉不同的部分，batch指的是一次输入多少张图片进行一次迭代，batch如果设置的太大，可能导致内存爆掉，提示核心已转存储，这时候你就需要调整一下batch的大小，最好为2的倍数。笔者2G的GTX960m的显卡上设置为16。训练完进行测试的时候需要将这里再改回test的batch与subdivisions，即batch=1,subdivisions=1。
- 设置classes=1
- 设置filters=(classes + 5) * 3，filters的设置依据可以看这里
下面提供一个我自己的配置文件，需要修改的地方已经用星号标出，如果对参数修改还有疑问，可以参考这篇博客 YOLOv3: 训练自己的数据
[net]
# Testing ### 测试模式 ★★★
# batch=1
# subdivisions=1
# Training ### 训练模式，每次前向的图片数目 = batch/subdivisions ★★★
batch=32
subdivisions=8
width=416 ### 网络的输入宽、高、通道数 ★★
height=416
channels=3
momentum=0.9 ### 动量 ★
decay=0.0005 ### 权重衰减 ★
angle=0
saturation = 1.5 ### 饱和度 ★
exposure = 1.5 ### 曝光度 ★
hue=.1 ### 色调 ★
learning_rate=0.001 ### 学习率 ★★★
burn_in=1000 ### 学习率控制的参数
max_batches = 50200 ### 迭代次数 ★★★
policy=steps ### 学习率策略 ★★
steps=40000,45000 ### 学习率变动步长 ★★
scales=.1,.1 ### 学习率变动因子 ★★
[convolutional]
batch_normalize=1 ### BN
filters=32 ### 卷积核数目
size=3 ### 卷积核尺寸
stride=1 ### 卷积核步长
pad=1 ### pad
activation=leaky ### 激活函数
# Downsample
[convolutional]
batch_normalize=1
filters=64
size=3
stride=2
pad=1
activation=leaky
[convolutional]
batch_normalize=1
filters=32
size=1
stride=1
pad=1
activation=leaky
[convolutional]
batch_normalize=1
filters=64
size=3
stride=1
pad=1
activation=leaky
[shortcut]
from=-3
activation=linear
# Downsample
[convolutional]
batch_normalize=1
filters=128
size=3
stride=2
pad=1
activation=leaky
[convolutional]
batch_normalize=1
filters=64
size=1
stride=1
pad=1
activation=leaky
[convolutional]
batch_normalize=1
filters=128
size=3
stride=1
pad=1
activation=leaky
[shortcut]
from=-3 ### 连接层
activation=linear
[convolutional]
batch_normalize=1
filters=64
size=1
stride=1
pad=1
activation=leaky
[convolutional]
batch_normalize=1
filters=128
size=3
stride=1
pad=1
activation=leaky
[shortcut]
from=-3
activation=linear
# Downsample
[convolutional]
batch_normalize=1
filters=256
size=3
stride=2
pad=1
activation=leaky
[convolutional]
batch_normalize=1
filters=128
size=1
stride=1
pad=1
activation=leaky
[convolutional]
batch_normalize=1
filters=256
size=3
stride=1
pad=1
activation=leaky
[shortcut]
from=-3
activation=linear
[convolutional]
batch_normalize=1
filters=128
size=1
stride=1
pad=1
activation=leaky
[convolutional]
batch_normalize=1
filters=256
size=3
stride=1
pad=1
activation=leaky
[shortcut]
from=-3
activation=linear
[convolutional]
batch_normalize=1
filters=128
size=1
stride=1
pad=1
activation=leaky
[convolutional]
batch_normalize=1
filters=256
size=3
stride=1
pad=1
activation=leaky
[shortcut]
from=-3
activation=linear
[convolutional]
batch_normalize=1
filters=128
size=1
stride=1
pad=1
activation=leaky
[convolutional]
batch_normalize=1
filters=256
size=3
stride=1
pad=1
activation=leaky
[shortcut]
from=-3
activation=linear
[convolutional]
batch_normalize=1
filters=128
size=1
stride=1
pad=1
activation=leaky
[convolutional]
batch_normalize=1
filters=256
size=3
stride=1
pad=1
activation=leaky
[shortcut]
from=-3
activation=linear
[convolutional]
batch_normalize=1
filters=128
size=1
stride=1
pad=1
activation=leaky
[convolutional]
batch_normalize=1
filters=256
size=3
stride=1
pad=1
activation=leaky
[shortcut]
from=-3
activation=linear
[convolutional]
batch_normalize=1
filters=128
size=1
stride=1
pad=1
activation=leaky
[convolutional]
batch_normalize=1
filters=256
size=3
stride=1
pad=1
activation=leaky
[shortcut]
from=-3
activation=linear
[convolutional]
batch_normalize=1
filters=128
size=1
stride=1
pad=1
activation=leaky
[convolutional]
batch_normalize=1
filters=256
size=3
stride=1
pad=1
activation=leaky
[shortcut]
from=-3
activation=linear
# Downsample
[convolutional]
batch_normalize=1
filters=512
size=3
stride=2
pad=1
activation=leaky
[convolutional]
batch_normalize=1
filters=256
size=1
stride=1
pad=1
activation=leaky
[convolutional]
batch_normalize=1
filters=512
size=3
stride=1
pad=1
activation=leaky
[shortcut]
from=-3
activation=linear
[convolutional]
batch_normalize=1
filters=256
size=1
stride=1
pad=1
activation=leaky
[convolutional]
batch_normalize=1
filters=512
size=3
stride=1
pad=1
activation=leaky
[shortcut]
from=-3
activation=linear
[convolutional]
batch_normalize=1
filters=256
size=1
stride=1
pad=1
activation=leaky
[convolutional]
batch_normalize=1
filters=512
size=3
stride=1
pad=1
activation=leaky
[shortcut]
from=-3
activation=linear
[convolutional]
batch_normalize=1
filters=256
size=1
stride=1
pad=1
activation=leaky
[convolutional]
batch_normalize=1
filters=512
size=3
stride=1
pad=1
activation=leaky
[shortcut]
from=-3
activation=linear
[convolutional]
batch_normalize=1
filters=256
size=1
stride=1
pad=1
activation=leaky
[convolutional]
batch_normalize=1
filters=512
size=3
stride=1
pad=1
activation=leaky
[shortcut]
from=-3
activation=linear
[convolutional]
batch_normalize=1
filters=256
size=1
stride=1
pad=1
activation=leaky
[convolutional]
batch_normalize=1
filters=512
size=3
stride=1
pad=1
activation=leaky
[shortcut]
from=-3
activation=linear
[convolutional]
batch_normalize=1
filters=256
size=1
stride=1
pad=1
activation=leaky
[convolutional]
batch_normalize=1
filters=512
size=3
stride=1
pad=1
activation=leaky
[shortcut]
from=-3
activation=linear
[convolutional]
batch_normalize=1
filters=256
size=1
stride=1
pad=1
activation=leaky
[convolutional]
batch_normalize=1
filters=512
size=3
stride=1
pad=1
activation=leaky
[shortcut]
from=-3
activation=linear
# Downsample
[convolutional]
batch_normalize=1
filters=1024
size=3
stride=2
pad=1
activation=leaky
[convolutional]
batch_normalize=1
filters=512
size=1
stride=1
pad=1
activation=leaky
[convolutional]
batch_normalize=1
filters=1024
size=3
stride=1
pad=1
activation=leaky
[shortcut]
from=-3
activation=linear
[convolutional]
batch_normalize=1
filters=512
size=1
stride=1
pad=1
activation=leaky
[convolutional]
batch_normalize=1
filters=1024
size=3
stride=1
pad=1
activation=leaky
[shortcut]
from=-3
activation=linear
[convolutional]
batch_normalize=1
filters=512
size=1
stride=1
pad=1
activation=leaky
[convolutional]
batch_normalize=1
filters=1024
size=3
stride=1
pad=1
activation=leaky
[shortcut]
from=-3
activation=linear
[convolutional]
batch_normalize=1
filters=512
size=1
stride=1
pad=1
activation=leaky
[convolutional]
batch_normalize=1
filters=1024
size=3
stride=1
pad=1
activation=leaky
[shortcut]
from=-3
activation=linear
######################
[convolutional]
batch_normalize=1
filters=512
size=1
stride=1
pad=1
activation=leaky
[convolutional]
batch_normalize=1
size=3
stride=1
pad=1
filters=1024
activation=leaky
[convolutional]
batch_normalize=1
filters=512
size=1
stride=1
pad=1
activation=leaky
[convolutional]
batch_normalize=1
size=3
stride=1
pad=1
filters=1024
activation=leaky
[convolutional]
batch_normalize=1
filters=512
size=1
stride=1
pad=1
activation=leaky
[convolutional]
batch_normalize=1
size=3
stride=1
pad=1
filters=1024
activation=leaky
[convolutional]
size=1
stride=1
pad=1
filters=18 ★★★
activation=linear
[yolo]
mask = 6,7,8
anchors = 10,13, 16,30, 33,23, 30,61, 62,45, 59,119, 116,90, 156,198, 373,326
classes=1 ★★★
num=9
jitter=.3
ignore_thresh = .5
truth_thresh = 1
random=0 ★★★
[route]
layers = -4
[convolutional]
batch_normalize=1
filters=256
size=1
stride=1
pad=1
activation=leaky
[upsample]
stride=2
[route]
layers = -1, 61
[convolutional]
batch_normalize=1
filters=256
size=1
stride=1
pad=1
activation=leaky
[convolutional]
batch_normalize=1
size=3
stride=1
pad=1
filters=512
activation=leaky
[convolutional]
batch_normalize=1
filters=256
size=1
stride=1
pad=1
activation=leaky
[convolutional]
batch_normalize=1
size=3
stride=1
pad=1
filters=512
activation=leaky
[convolutional]
batch_normalize=1
filters=256
size=1
stride=1
pad=1
activation=leaky
[convolutional]
batch_normalize=1
size=3
stride=1
pad=1
filters=512
activation=leaky
[convolutional]
size=1
stride=1
pad=1
filters=18 ★★★
activation=linear
[yolo]
mask = 3,4,5
anchors = 10,13, 16,30, 33,23, 30,61, 62,45, 59,119, 116,90, 156,198, 373,326
classes=1 ★★★
num=9
jitter=.3
ignore_thresh = .5
truth_thresh = 1
random=0 ★★★
[route]
layers = -4
[convolutional]
batch_normalize=1
filters=128
size=1
stride=1
pad=1
activation=leaky
[upsample]
stride=2
[route]
layers = -1, 36
[convolutional]
batch_normalize=1
filters=128
size=1
stride=1
pad=1
activation=leaky
[convolutional]
batch_normalize=1
size=3
stride=1
pad=1
filters=256
activation=leaky
[convolutional]
batch_normalize=1
filters=128
size=1
stride=1
pad=1
activation=leaky
[convolutional]
batch_normalize=1
size=3
stride=1
pad=1
filters=256
activation=leaky
[convolutional]
batch_normalize=1
filters=128
size=1
stride=1
pad=1
activation=leaky
[convolutional]
batch_normalize=1
size=3
stride=1
pad=1
filters=256
activation=leaky
[convolutional]
size=1
stride=1
pad=1
filters=18 ### 3x(classes + 4coor + 1prob) = 3x(20+4+1) = 75 ★★★
activation=linear
[yolo]
mask = 0,1,2 ### mask序号 ★
anchors = 10,13, 16,30, 33,23, 30,61, 62,45, 59,119, 116,90, 156,198, 373,326 ★★★
classes=20 ### 类比数目 ★★★
num=9
jitter=.3 ### 数据扩充的抖动操作 ★
ignore_thresh = .5 ### 文章中的阈值1 ★
truth_thresh = 1 ### 文章中的阈值2 ★
random=0 ### 多尺度训练开关 ★★★
七、训练过程中参数的保存
在linux下可通过数据重定向，将训练过程中输出的数据保存到文件中，并通过python脚本解析可视化，具体的步骤可参考博客YOLOv3使用笔记——曲线可视化文中已经写得很详细，并提供了可视化的脚本。
八、批量测试图片
总结
到这里就完成了整个YOLOv3从数据集准备到得到训练后权重文件的过程，并且还提供了训练过程的可视化及批量图片测试，应该能满足大部分场景下的需求了，在这里还是要感谢以上博主的无私分享，最后建议大家有时间可以看一下源码，如果读懂了源码，以上这些就都不是问题了。