tensorflow+ssd_mobilenet實現目標檢測的訓練

本文在Ubuntu下使用tensorflow的object detection API來訓練自己的數據集。所用模型為ssd_mobilenet，也可以使用其他的模型。當然也可以在windows下訓練，代碼上沒有多大差別，主要是配置環境那里，比較麻煩（windows和linux下都一樣麻煩）。

一、配置環境

1. 在GitHub上下載所需的models文件，地址：https://github.com/tensorflow/models。

2. 安裝pillow、lxml。Jupyter Notebook和matplotlib這兩個我就不說了，肯定早就裝好了。

pip install pillow
pip install lxml

3. 編譯protobuf，object detection API是使用protobuf來訓練模型和配置參數，所以得先編譯protobuf，下載地址：https://github.com/google/protobuf/releases

windows的配置參考：https://blog.csdn.net/dy_guox/article/details/79081499

Ubuntu下的配置為：

下載地址 https://github.com/google/protobuf/releases
下載protobuf-all-3.7.0.tar.gz

命令如下：

tar zxvf protobuf-all-3.7.0.tar.gz 
cd protobuf-3.7.0/
./autogen.sh 
./configure --prefix=/usr/local/protobuf
make

make check 出現錯誤，不過沒事

修改配置文件

（1）vim /etc/profile，添加

　　　　　export PATH=$PATH:/usr/local/protobuf/bin/
　　　　　export PKG_CONFIG_PATH=/usr/local/protobuf/lib/pkgconfig/

　　　　保存執行source /etc/profile
　　　　vim ~/.profile，添加
　　　　　　export PATH=$PATH:/usr/local/protobuf/bin/
　　　　　　export PKG_CONFIG_PATH=/usr/local/protobuf/lib/pkgconfig/ 否則會出現登陸用戶找不到protoc命令
（2）配置動態鏈接庫
　　　　vim /etc/ld.so.conf 在新的一行添加
　　　　/usr/local/protobuf/lib
　　　　執行命令 sudo ldconfig

安裝完成

執行protoc --version,會出現當前libporoto的版本信息
libprotoc 2.6.1

 

二、在Ubuntu內先制作好自己的VOC數據集。制作過程如下：

在Ubuntu內制作自己的VOC數據集

 

三、將樣本數據轉換為TFRecord格式

1. 新建 train_test_split.py 把xml文件數據集分為了train、test、validation三部分，並存儲在Annotations文件夾中，訓練驗證集占80%，測試集占20%。訓練集占訓練驗證集的80%。代碼如下：

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))

 

 

2. 把xml轉換成csv文件，xml_to_csv.py 將生成的csv文件放在 object_detection/data/

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/trafficlight_{}_labels.csv'.format(directory), index=None)  
        print('Successfully converted xml to csv.')

main()  

 

3. 生成tfrecord文件，generate_tfrecord.py

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Created on Tue Mar  5 15:28:55 2019

@author: z
"""

"""
Usage:
  # From tensorflow/models/
  # Create train data:
  python generate_tfrecord.py --csv_input=data/tv_vehicle_labels.csv  --output_path=train.record
  # Create test data:
  python generate_tfrecord.py --csv_input=data/test_labels.csv  --output_path=test.record
"""



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/z/work/models-master/research/')

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):
        # 你的所有類別
    if row_label == 'red pedestrian':
            return 1
    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.gfile.GFile(os.path.join(path, '{}'.format(group.filename)), 'rb') as fid:
        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.python_io.TFRecordWriter(FLAGS.output_path)
    path = os.path.join(os.getcwd(), 'object_detection/images/voc/VOCdevkit/VOC2018/JPEGImages/')
    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個轉換，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()

主要是在 row_label 這里要添加上你標注的類別，字符串 row_label 應於labelImg中標注的名稱相同。同樣 path 為圖片的路徑。

cd /home/z/work/models-master/research

python generate_tfrecord.py --csv_input=object_detection/data/trafficlight_train_labels.csv --output_path=object_detection/data/

trafficlight_train.tfrecord

generate_tfrecord.py 需要在research目錄下，也就是object_detection的上級目錄，因為在腳本中使用了 object_detection.utils，如果在 object_detection 下執行命令會報錯（No module named object_detection）。

其實這句命令很好理解，其實就是根據腳本中提供的圖片路徑，找到圖片所在。至於是哪些圖片？由csv文件來決定。csv文件主要就是記錄圖片的名稱、類別、以及標記框的坐標。如下圖所示：

類似的，我們可以輸入如下命令，將驗證集和測試集也轉換為tfrecord格式。

python generate_tfrecord.py --csv_input=object_detection/data/trafficlight_validation_labels.csv --output_path=object_detection/data/
trafficlight_validation.tfrecord

python generate_tfrecord.py --csv_input=object_detection/data/trafficlight_test_labels.csv --output_path=object_detection/data/
trafficlight_test.tfrecord

 

四、訓練

1. 在object_detection/data文件夾下創建標簽分類的配置文件（label_map.txt），需要檢測幾種目標，將創建幾個id，代碼如下：

item {
  id: 1    # id 從1開始編號
  name: 'red pedestrian'
}

item {
  id: 2
  name: 'green pedestrian'
}

2. 配置管道配置文件，找到object_detection/samples/config/ssd_mobilenet_v1_coco.config，復制到data文件夾下。修改后的代碼如下：

# SSD with Mobilenet v1 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: 17
    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_v1'
      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: 0
      }
      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
    }
  }

  # 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: "data/trafficlight_train.tfrecord"
  }
  label_map_path: "data/trafficlight_label_map.pbtxt"
}

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: "data/trafficlight_validation.tfrecord"
  }
  label_map_path: "data/trafficlight_label_map.pbtxt"
  shuffle: false
  num_readers: 1
}

可以修改的有好幾處，在代碼中都標紅了。建議還是將配置文件看一下。主要包括：標注的類別數目、batch_size(建議設置小一點)、學習率和退化率、訓練的總步數、訓練集和驗證集的tfrecord的路徑、label_map的路徑

3. 開始訓練

python train.py --logtostderr --train_dir=training/ --pipeline_config_path=data/ssd_mobilenet_v1_coco.config

 參考地址：

https://www.cnblogs.com/White-xzx/p/9503203.html