Tensorflow官方模型库
升级到最新的Tensorflow2
pip install tf-nightly
pip 自动安装所有的模型和依赖项
pip install tf-models-official
若要安装最新的更改则:
pip install tf-models-nightly
git clone https://github.com/tensorflow/models.git
git下载时出现
error: RPC failed; curl 56 GnuTLS recv error (-54): Error in the pull function.
fatal: The remote end hung up unexpectedly
fatal: 过早的文件结束符(EOF)
fatal: index-pack failed
先输入:
git init
然后再
git config http.postBuffer 524288000
export PYTHONPATH=$PYTHONPATH:/path/to/models
pip install --user -r official/requirements.txt
也可以一个一个安装
其他依赖库参考:requirement.txt
six
google-api-python-client>=1.6.7
google-cloud-bigquery>=0.31.0
kaggle>=1.3.9
numpy>=1.15.4
oauth2client
pandas>=0.22.0
psutil>=5.4.3
py-cpuinfo>=3.3.0
scipy>=0.19.1
tensorflow-hub>=0.6.0
tensorflow-model-optimization>=0.4.1
tensorflow-datasets
tensorflow-addons
dataclasses
gin-config
tf_slim>=1.1.0
Cython
matplotlib
pyyaml>=5.1
#CV related dependencies
opencv-python-headless
Pillow
pycocotools
#NLP related dependencies
seqeval
sentencepiece
笔者在Ubuntu16.04的服务器进行环境配置
conda create -n tfssd python=3.6
conda activate tfssd
conda install tensorflow-gpu=1.15.0
pip install lxml, Cython, contextlib2 , jupyter, matplotlib, pillow
git clone https://github.com/tensorflow/models.git
或者直接下载安装包,进行解压
git clone https://github.com/cocodataset/cocoapi.git
cd cocoapi/PythonAPI
make
make install
python setup.py install
cp -r ./cocoapi/PythonAPI/pycocotools ./models-master/research/ # 复制pycocotools到models-master/research/
protobuf下载地址:https://github.com/google/protobuf/releases
笔者下载的是protoc-3.13.0-linux-x86_64.zip,将压缩包进行解压。
cd /models-master/protoc-3.13.0-linux-x86_64/bin # 进入bin文件夹
chmod +x configure # configure没有执行权限,通过chmod给其加上x权限
export PATH=/opt/protoc-3.6.1-linux-x86_64/bin:$PATH # 增加环境变量
source ~/.bashrc
测试是否安装成功:
cd ~/models-master/research # 进入tensorflow的项目文件
protoc object_detection/protos/*proto --python_out=.
protoc --version
输出 libprotoc 3.13.0
下载VOC数据集,放在 model-master/research/object_detection/images/
(images是笔者新建的文件夹用于储存原始数据集)
VOC数据集有五个文件夹
├── Annotations # 存放xml文件,主要是记录标记框位置信息
├── ImageSets # 存放的都是txt文件,txt文件中每一行包含一个图片的名称,末尾会加上+1或者-1表示正负样本
├── Action
├── Layout
├── Main
└── Segmentation
├── JPEGImages # 存放源图片
├── SegmentationClass
└── SegmentationObject
制作自己的数据集时只需要用到Annotations、ImageSets、JPEGImages三个文件夹
具体制作过程见【VOC数据集】制作
├──hengfeng # 笔者的目标名
├── Annotations # 存放xml文件,主要是记录标记框位置信息
├── JPEGImages # 存放源图片
新建train_test_split.py
把xml文件数据集分为了train、test、validation三部分,并存储在Annotations
文件夹中,训练验证集占80%,测试集占20%。训练集占训练验证集的80%。代码如下:
import os
import random
import time
import shutil
xmlfilepath=r'./Annotations'
saveBasePath=r"./Annotations"
trainval_percent=0.8
train_percent=0.8
total_xml = os.listdir(xmlfilepath)
num=len(total_xml)
list=range(num)
tv=int(num*trainval_percent)
tr=int(tv*train_percent)
trainval= random.sample(list,tv)
train=random.sample(trainval,tr)
print("train and val size",tv)
print("train size",tr)
start = time.time()
test_num=0
val_num=0
train_num=0
for i in list:
name=total_xml[i]
if i in trainval: #train and val set
if i in train:
directory="train"
train_num += 1
xml_path = os.path.join(os.getcwd(), 'Annotations/{}'.format(directory))
if(not os.path.exists(xml_path)):
os.mkdir(xml_path)
filePath=os.path.join(xmlfilepath,name)
newfile=os.path.join(saveBasePath,os.path.join(directory,name))
shutil.copyfile(filePath, newfile)
else:
directory="validation"
xml_path = os.path.join(os.getcwd(), 'Annotations/{}'.format(directory))
if(not os.path.exists(xml_path)):
os.mkdir(xml_path)
val_num += 1
filePath=os.path.join(xmlfilepath,name)
newfile=os.path.join(saveBasePath,os.path.join(directory,name))
shutil.copyfile(filePath, newfile)
else:
directory="test"
xml_path = os.path.join(os.getcwd(), 'Annotations/{}'.format(directory))
if(not os.path.exists(xml_path)):
os.mkdir(xml_path)
test_num += 1
filePath=os.path.join(xmlfilepath,name)
newfile=os.path.join(saveBasePath,os.path.join(directory,name))
shutil.copyfile(filePath, newfile)
end = time.time()
seconds=end-start
print("train total : "+str(train_num))
print("validation total : "+str(val_num))
print("test total : "+str(test_num))
total_num=train_num+val_num+test_num
print("total number : "+str(total_num))
print( "Time taken : {0} seconds".format(seconds))
xml_to_csv.py
将生成的csv文件放在 model-master/research/object_detection/data/hengfeng/
代码如下:
import os
import glob
import pandas as pd
import xml.etree.ElementTree as ET
def xml_to_csv(path):
xml_list = []
for xml_file in glob.glob(path + '/*.xml'):
tree = ET.parse(xml_file)
root = tree.getroot()
print(root.find('filename').text)
for member in root.findall('object'):
value = (root.find('filename').text,
int(root.find('size')[0].text), #width
int(root.find('size')[1].text), #height
member[0].text,
int(member[4][0].text),
int(float(member[4][1].text)),
int(member[4][2].text),
int(member[4][3].text)
)
xml_list.append(value)
column_name = ['filename', 'width', 'height', 'class', 'xmin', 'ymin', 'xmax', 'ymax']
xml_df = pd.DataFrame(xml_list, columns=column_name)
return xml_df
def main():
for directory in ['train','test','validation']:
xml_path = os.path.join(os.getcwd(), 'Annotations/{}'.format(directory))
xml_df = xml_to_csv(xml_path)
# xml_df.to_csv('whsyxt.csv', index=None)
xml_df.to_csv('/home/z/work/models-master/research/object_detection/data/hengfeng/hengfeng_{}_labels.csv'.format(directory), index=None)
print('Successfully converted xml to csv.')
main()
generate_tfrecord.py
代码如下
# -*- coding: utf-8 -*-
"""
2020.09.23:alian, 最终可使用版本*********
生成tfrecord文件
"""
"""
Usage:
# Create train data:
python generate_tfrecord.py --csv_input=object_detection/data/hengfeng_train_labels.csv --output_path=object_detection/data/hengfeng_train.tfrecord
# Create val data:
python generate_tfrecord.py --csv_input=object_detection/data/hengfeng_validation_labels.csv --output_path=object_detection/data/hengfeng_validation.tfrecord
# Create test data:
python generate_tfrecord.py --csv_input=object_detection/data/hengfeng_test_labels.csv --output_path=object_detection/data/hengfeng_test.tfrecord
"""
import os
import io
import pandas as pd
import tensorflow as tf
from PIL import Image
from object_detection.utils import dataset_util
from collections import namedtuple, OrderedDict
os.chdir('/home/lianlirong/models-master/research/') # 改变当前目录到指定的目录
flags = tf.app.flags
flags.DEFINE_string('csv_input', '', 'Path to the CSV input')
flags.DEFINE_string('output_path', '', 'Path to output TFRecord')
FLAGS = flags.FLAGS
# TO-DO replace this with label map
def class_text_to_int(row_label):
# 你的所有类别——————————————————————————————————————————————————————————
if row_label == 'Hengfeng':
return 1
elif row_label == 'Beijing':
return 2
else:
return None
def split(df, group):
data = namedtuple('data', ['filename', 'object'])
gb = df.groupby(group)
return [data(filename, gb.get_group(x)) for filename, x in zip(gb.groups.keys(), gb.groups)]
def create_tf_example(group, path):
with tf.io.gfile.GFile(os.path.join(path, '{}'.format(group.filename)), 'rb') as fid: # tf.gfile.GFile 替换成 tf.io.gfile.GFile
encoded_jpg = fid.read()
encoded_jpg_io = io.BytesIO(encoded_jpg)
image = Image.open(encoded_jpg_io)
width, height = image.size
filename = group.filename.encode('utf8')
image_format = b'jpg'
xmins = []
xmaxs = []
ymins = []
ymaxs = []
classes_text = []
classes = []
for index, row in group.object.iterrows():
xmins.append(row['xmin'] / width)
xmaxs.append(row['xmax'] / width)
ymins.append(row['ymin'] / height)
ymaxs.append(row['ymax'] / height)
classes_text.append(row['class'].encode('utf8'))
classes.append(class_text_to_int(row['class']))
tf_example = tf.train.Example(features=tf.train.Features(feature={
'image/height': dataset_util.int64_feature(height),
'image/width': dataset_util.int64_feature(width),
'image/filename': dataset_util.bytes_feature(filename),
'image/source_id': dataset_util.bytes_feature(filename),
'image/encoded': dataset_util.bytes_feature(encoded_jpg),
'image/format': dataset_util.bytes_feature(image_format),
'image/object/bbox/xmin': dataset_util.float_list_feature(xmins),
'image/object/bbox/xmax': dataset_util.float_list_feature(xmaxs),
'image/object/bbox/ymin': dataset_util.float_list_feature(ymins),
'image/object/bbox/ymax': dataset_util.float_list_feature(ymaxs),
'image/object/class/text': dataset_util.bytes_list_feature(classes_text),
'image/object/class/label': dataset_util.int64_list_feature(classes),
}))
return tf_example
def main(_):
writer = tf.io.TFRecordWriter(FLAGS.output_path)
path = os.path.join(os.getcwd(), './models-master/research/object_detection/images/VOC2007-hengfeng/JPEGImages/') # 原图所在的路径
examples = pd.read_csv(FLAGS.csv_input)
grouped = split(examples, 'filename')
num = 0
for group in grouped:
num += 1
tf_example = create_tf_example(group, path)
writer.write(tf_example.SerializeToString())
if (num % 100 == 0): # 每完成100个转换,打印一次
print(num)
writer.close()
output_path = os.path.join(os.getcwd(), FLAGS.output_path)
print('Successfully created the TFRecords: {}'.format(output_path))
if __name__ == '__main__':
tf.app.run()
# tf.compat.v1.app.run
主要是在 row_label 这里要添加上你标注的类别,字符串 row_label 应于labelImg中标注的名称相同;同样 path 为图片的路径。
cd /models-master/research
python generate_tfrecord.py --csv_input=object_detection/data/hengfeng/hengfeng_train_labels.csv --output_path=object_detection/data/hengfeng/hengfeng_train.tfrecord
generate_tfrecord.py
需要在research
目录下,也就是object_detection
的上级目录,因为在脚本中使用了 object_detection.utils,如果在 object_detection 下执行命令会报错(No module named object_detection)。
其实这句命令很好理解,其实就是根据脚本中提供的图片路径,找到图片所在。至于是哪些图片?由csv文件来决定。csv文件主要就是记录图片的名称、类别、以及标记框的坐标。如下图所示:
类似的,我们可以输入如下命令,将验证集和测试集也转换为tfrecord格式。
python generate_tfrecord.py --csv_input=object_detection/data/hengfeng/hengfeng_validation_labels.csv --output_path=object_detection/data/hengfeng/hengfeng_validation.tfrecord
python generate_tfrecord.py --csv_input=object_detection/data/hengfeng/hengfeng_test_labels.csv --output_path=object_detection/data/hengfeng/hengfeng_test.tfrecord
item {
id: 1 # id 从1开始编号
name: 'Hengfeng'
}
item {
id: 2
name: 'Beijing'
}
object_detection/samples/config/ssd_mobilenet_v2_coco.config
,复制到data/object/文件夹下。修改后的代码如下:# SSD with Mobilenet v2 configuration for MSCOCO Dataset.
# Users should configure the fine_tune_checkpoint field in the train config as
# well as the label_map_path and input_path fields in the train_input_reader and
# eval_input_reader. Search for "PATH_TO_BE_CONFIGURED" to find the fields that
# should be configured.
model {
ssd {
num_classes: 2 ######
box_coder {
faster_rcnn_box_coder {
y_scale: 10.0
x_scale: 10.0
height_scale: 5.0
width_scale: 5.0
}
}
matcher {
argmax_matcher {
matched_threshold: 0.5
unmatched_threshold: 0.5
ignore_thresholds: false
negatives_lower_than_unmatched: true
force_match_for_each_row: true
}
}
similarity_calculator {
iou_similarity {
}
}
anchor_generator {
ssd_anchor_generator {
num_layers: 6
min_scale: 0.2
max_scale: 0.95
aspect_ratios: 1.0
aspect_ratios: 2.0
aspect_ratios: 0.5
aspect_ratios: 3.0
aspect_ratios: 0.3333
}
}
image_resizer {
fixed_shape_resizer {
height: 300
width: 300
}
}
box_predictor {
convolutional_box_predictor {
min_depth: 0
max_depth: 0
num_layers_before_predictor: 0
use_dropout: false
dropout_keep_probability: 0.8
kernel_size: 1
box_code_size: 4
apply_sigmoid_to_scores: false
conv_hyperparams {
activation: RELU_6,
regularizer {
l2_regularizer {
weight: 0.00004
}
}
initializer {
truncated_normal_initializer {
stddev: 0.03
mean: 0.0
}
}
batch_norm {
train: true,
scale: true,
center: true,
decay: 0.9997,
epsilon: 0.001,
}
}
}
}
feature_extractor {
type: 'ssd_mobilenet_v2'
min_depth: 16
depth_multiplier: 1.0
conv_hyperparams {
activation: RELU_6,
regularizer {
l2_regularizer {
weight: 0.00004
}
}
initializer {
truncated_normal_initializer {
stddev: 0.03
mean: 0.0
}
}
batch_norm {
train: true,
scale: true,
center: true,
decay: 0.9997,
epsilon: 0.001,
}
}
}
loss {
classification_loss {
weighted_sigmoid {
}
}
localization_loss {
weighted_smooth_l1 {
}
}
hard_example_miner {
num_hard_examples: 3000
iou_threshold: 0.99
loss_type: CLASSIFICATION
max_negatives_per_positive: 3
min_negatives_per_image: 3
}
classification_weight: 1.0
localization_weight: 1.0
}
normalize_loss_by_num_matches: true
post_processing {
batch_non_max_suppression {
score_threshold: 1e-8
iou_threshold: 0.6
max_detections_per_class: 100
max_total_detections: 100
}
score_converter: SIGMOID
}
}
}
train_config: {
batch_size: 2 ######
optimizer {
rms_prop_optimizer: {
learning_rate: {
exponential_decay_learning_rate {
initial_learning_rate: 0.00001 ######
decay_steps: 800720
decay_factor: 0.95 ######
}
}
momentum_optimizer_value: 0.9
decay: 0.9
epsilon: 1.0
}
}
fine_tune_checkpoint: "/models-master/research/object_detection/ssd_mobilenet_v2_coco_2018_03_29/model.ckpt"
fine_tune_checkpoint_type: "detection"
# Note: The below line limits the training process to 200K steps, which we
# empirically found to be sufficient enough to train the pets dataset. This
# effectively bypasses the learning rate schedule (the learning rate will
# never decay). Remove the below line to train indefinitely.
num_steps: 6000 #######
data_augmentation_options {
random_horizontal_flip {
}
}
data_augmentation_options {
ssd_random_crop {
}
}
}
train_input_reader: {
tf_record_input_reader {
input_path: "/models-master/research/object_detection/data/hengfeng/hengfeng_train.tfrecord" ###### 建议使用绝对路径
}
label_map_path: "/models-master/research/object_detection/data/hengfeng/hengfeng_label_map.pbtxt" ###### 建议使用绝对路径
}
eval_config: {
num_examples: 8000
# Note: The below line limits the evaluation process to 10 evaluations.
# Remove the below line to evaluate indefinitely.
max_evals: 10
}
eval_input_reader: {
tf_record_input_reader {
input_path: "/models-master/research/object_detection/data/hengfeng/hengfeng_validation.tfrecord" ###### 建议使用绝对路径
}
label_map_path: "/models-master/research/object_detection/data/hengfeng/hengfeng_label_map.pbtxt" ###### 建议使用绝对路径
shuffle: false
num_readers: 1
}
修改的部分主要包括:9
:标注的类别数目、141
:batch_size(建议设置小一点)、146
:学习率和退化率、156
:预训练权重的路径,162
:训练的总步数、175
;177
:训练集和189
;191
:验证集的tfrecord的路径、label_map的路径.
输入训练指令:
python object_detection/model_main.py --logtostderr --model_dir=/models-master/logs/ --pipeline_config_path=/models-master/research/object_detection/data/hengfeng/ssd_mobilenet_v2_coco.config
训练时终端显示如下:
训练过程就到此结束,后续的测试将在下一篇博客进行说明。
下面整理下以上提到文件的放置位置:(以下的hengfeng
为笔者的训练目标)
├──models-master (tensorflow项目文件)
├── logs # 笔者存放训练模型的目录
├──hengfeng
├── research
├──generate_tfrecord.py # 生成tfrecord文件的代码
├──object_detection
├──ssd_mobilenet_v2_coco_2018_03_29
├──model.ckpt # 预训练模型文件
├──model_main.py # 训练代码文件
├──images
├──hengfeng # 目标数据集
├──Annotations # xml文件
├──JPEGImages # 原图片
├──train_test_split.py # 划分训练测试集的代码
├──xml_to_csv.py # 生成csv的代码文件
├──others_object …# 其他目标的数据集
├──data
├──hengfeng # 目标训练的必要文件
├──hengfeng_label_map.pbtxt # 目标标签文件
├──hengfeng_train.tfrecord #生成的tfrecord文件
├──hengfeng_val.tfrecord
├──hengfeng_test.tfrecord
├──hengfeng_train_labels.csv # 生成的csv文件
├──hengfeng_val_labels.csv
├──hengfeng_test_labels.csv
├──ssd_mobilenet_v2_coco.config # 训练的配置文件
├──others_object #其他目标