[Tensorflow] Object Detection API - build your training environment

时间：2017-09-23 17:21:04 收藏：0 阅读：2605

Prepare protoc

技术分享

Create folder: protoc and unzip it.

unsw@unsw-UX303UB$ ls
models  Others  protoc  train_data

unsw@unsw-UX303UB$ ls protoc/
bin  include  readme.txt

unsw@unsw-UX303UB$ ls protoc/bin/
protoc

Prepare model

Download model folder from tensorflow github.

unsw@unsw-UX303UB$ git clone https://github.com/tensorflow/models.git
Cloning into ‘models‘...
remote: Counting objects: 7518, done.
remote: Compressing objects: 100% (5/5), done.
remote: Total 7518 (delta 0), reused 1 (delta 0), pack-reused 7513
Receiving objects: 100% (7518/7518), 157.87 MiB | 1.17 MiB/s, done.
Resolving deltas: 100% (4053/4053), done.
Checking connectivity... done.

unsw@unsw-UX303UB$ ls
annotations  images  models  Others  raccoon_labels.csv  xml_to_csv.py

unsw@unsw-UX303UB$ ls models/
AUTHORS     CONTRIBUTING.md    LICENSE   README.md  tutorials
CODEOWNERS  ISSUE_TEMPLATE.md  official  research   WORKSPACE

Enter: models/research/

# Set python env.
$ export PYTHONPATH=/home/unsw/Dropbox/Programmer/1-python/Tensorflow/ssd_proj/models/research/slim:$PYTHONPATH

$ python object_detection/builders/model_builder_test.py
.......
----------------------------------------------------------------------
Ran 7 tests in 0.022s

OK

Prepare train.record

Download: https://github.com/datitran/raccoon_dataset/blob/master/generate_tfrecord.py

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

  # Create test data:
  python generate_tfrecord.py --csv_input=data/test_labels.csv  --output_path=test.record
"""
from __future__ import division
from __future__ import print_function
from __future__ import absolute_import

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

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):
    if row_label == ‘raccoon‘:
        return 1
    else:
        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(), ‘images‘)
    examples = pd.read_csv(FLAGS.csv_input)
    grouped = split(examples, ‘filename‘)
    for group in grouped:
        tf_example = create_tf_example(group, path)
        writer.write(tf_example.SerializeToString())

    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()

generate_tfrecord.py

NB: we will do everything in models/research/ where the env has been set well.

So, move data/images here for generate_tfrecord.py

unsw@unsw-UX303UB$ pwd
/home/unsw/Dropbox/Programmer/1-python/Tensorflow/ssd_proj/models/research

unsw@unsw-UX303UB$ python ../../generate_tfrecord.py --csv_input=../../data/raccoon_labels.csv  --output_path=../../data/train.record
Successfully created the TFRecords: /home/unsw/Programmer/1-python/Tensorflow/ssd_proj/models/research/../../data/train.record

Now, we have got train_labels.csv (name changed from raccoon_labels.csv) train.record.

Prepare pre-train model

Download pre-trained model: https://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/detection_model_zoo.md

技术分享

Download configure file for pre-trained model: https://github.com/tensorflow/models/tree/master/research/object_detection/samples/configs

技术分享

This configure is already in our model folder:

unsw@unsw-UX303UB$ pwd
/home/unsw/Programmer/1-python/Tensorflow/ssd_proj/models/research/object_detection/samples/configs

unsw@unsw-UX303UB$ ls
faster_rcnn_inception_resnet_v2_atrous_coco.config  faster_rcnn_resnet101_voc07.config  faster_rcnn_resnet50_pets.config  ssd_inception_v2_pets.config
faster_rcnn_inception_resnet_v2_atrous_pets.config  faster_rcnn_resnet152_coco.config   rfcn_resnet101_coco.config        ssd_mobilenet_v1_coco.config
faster_rcnn_resnet101_coco.config                   faster_rcnn_resnet152_pets.config   rfcn_resnet101_pets.config        ssd_mobilenet_v1_pets.config
faster_rcnn_resnet101_pets.config                   faster_rcnn_resnet50_coco.config    ssd_inception_v2_coco.config

Configure based on your own data.

  1 # SSD with Mobilenet v1, configured for Oxford-IIIT Pets Dataset.
  2 # Users should configure the fine_tune_checkpoint field in the train config as
  3 # well as the label_map_path and input_path fields in the train_input_reader and
  4 # eval_input_reader. Search for "PATH_TO_BE_CONFIGURED" to find the fields that
  5 # should be configured.
  6 
  7 model {
  8   ssd {
  9     num_classes: 1



158   fine_tune_checkpoint: "ssd_mobilenet_v1_coco_11_06_2017/model.ckpt"
159   from_detection_checkpoint: true
160   # Note: The below line limits the training process to 200K steps, which we
161   # empirically found to be sufficient enough to train the pets dataset. This
162   # effectively bypasses the learning rate schedule (the learning rate will
163   # never decay). Remove the below line to train indefinitely.
164   num_steps: 200000
165   data_augmentation_options {
166     random_horizontal_flip {
167     }
168   }
169   data_augmentation_options {
170     ssd_random_crop {
171     }
172   }
173 }
174 
175 train_input_reader: {
176   tf_record_input_reader {
177     input_path: "data/train.record"
178   }
179   label_map_path: "data/object-detection.pbtxt"
180 }
181 
182 eval_config: {
183   num_examples: 2000
184   # Note: The below line limits the evaluation process to 10 evaluations.
185   # Remove the below line to evaluate indefinitely.
186   max_evals: 10
187 }
188 
189 eval_input_reader: {
190   tf_record_input_reader {
191     input_path: "data/test.record"
192   }
193   label_map_path: "data/object-detection.pbtxt"
194   shuffle: false
195   num_readers: 1
196 }

As above, we need to create object-detection.pbtxt as following:

item {
    id: 1
    name: ‘raccoon‘
}

Prepare training

Move all configure files based on ssd_mobilenet_v1_pets.config as following:

training folder: object-detection.pbtxt and ssd_mobilenet_v1_pets.config.

技术分享

data folder: train.record and train_labels.csv.

技术分享

Start training.

python object_detection/train.py --logtostderr --train_dir=training/ --pipeline_config_path=training/ssd_mobilenet_v1_pets.config

INFO:tensorflow:Starting Session.
INFO:tensorflow:Saving checkpoint to path training/model.ckpt
INFO:tensorflow:Starting Queues.
INFO:tensorflow:global_step/sec: 0
INFO:tensorflow:Recording summary at step 0.
INFO:tensorflow:global step 1: loss = 14.5804 (33.780 sec/step)
INFO:tensorflow:global step 2: loss = 12.6232 (19.210 sec/step)
INFO:tensorflow:global step 3: loss = 12.0996 (17.102 sec/step)

[Tensorflow] Object Detection API - build your training environment

Prepare protoc

Prepare model

Prepare train.record

Prepare pre-train model

Prepare training