File size: 9,800 Bytes
18ddfe2 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 |
# Copyright 2017 The TensorFlow Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# ==============================================================================
"""Tests for object_detection.trainer."""
import unittest
import tensorflow.compat.v1 as tf
import tf_slim as slim
from google.protobuf import text_format
from object_detection.core import losses
from object_detection.core import model
from object_detection.core import standard_fields as fields
from object_detection.legacy import trainer
from object_detection.protos import train_pb2
from object_detection.utils import tf_version
NUMBER_OF_CLASSES = 2
def get_input_function():
"""A function to get test inputs. Returns an image with one box."""
image = tf.random_uniform([32, 32, 3], dtype=tf.float32)
key = tf.constant('image_000000')
class_label = tf.random_uniform(
[1], minval=0, maxval=NUMBER_OF_CLASSES, dtype=tf.int32)
box_label = tf.random_uniform(
[1, 4], minval=0.4, maxval=0.6, dtype=tf.float32)
multiclass_scores = tf.random_uniform(
[1, NUMBER_OF_CLASSES], minval=0.4, maxval=0.6, dtype=tf.float32)
return {
fields.InputDataFields.image: image,
fields.InputDataFields.key: key,
fields.InputDataFields.groundtruth_classes: class_label,
fields.InputDataFields.groundtruth_boxes: box_label,
fields.InputDataFields.multiclass_scores: multiclass_scores
}
class FakeDetectionModel(model.DetectionModel):
"""A simple (and poor) DetectionModel for use in test."""
def __init__(self):
super(FakeDetectionModel, self).__init__(num_classes=NUMBER_OF_CLASSES)
self._classification_loss = losses.WeightedSigmoidClassificationLoss()
self._localization_loss = losses.WeightedSmoothL1LocalizationLoss()
def preprocess(self, inputs):
"""Input preprocessing, resizes images to 28x28.
Args:
inputs: a [batch, height_in, width_in, channels] float32 tensor
representing a batch of images with values between 0 and 255.0.
Returns:
preprocessed_inputs: a [batch, 28, 28, channels] float32 tensor.
true_image_shapes: int32 tensor of shape [batch, 3] where each row is
of the form [height, width, channels] indicating the shapes
of true images in the resized images, as resized images can be padded
with zeros.
"""
true_image_shapes = [inputs.shape[:-1].as_list()
for _ in range(inputs.shape[-1])]
return tf.image.resize_images(inputs, [28, 28]), true_image_shapes
def predict(self, preprocessed_inputs, true_image_shapes):
"""Prediction tensors from inputs tensor.
Args:
preprocessed_inputs: a [batch, 28, 28, channels] float32 tensor.
true_image_shapes: int32 tensor of shape [batch, 3] where each row is
of the form [height, width, channels] indicating the shapes
of true images in the resized images, as resized images can be padded
with zeros.
Returns:
prediction_dict: a dictionary holding prediction tensors to be
passed to the Loss or Postprocess functions.
"""
flattened_inputs = slim.flatten(preprocessed_inputs)
class_prediction = slim.fully_connected(flattened_inputs, self._num_classes)
box_prediction = slim.fully_connected(flattened_inputs, 4)
return {
'class_predictions_with_background': tf.reshape(
class_prediction, [-1, 1, self._num_classes]),
'box_encodings': tf.reshape(box_prediction, [-1, 1, 4])
}
def postprocess(self, prediction_dict, true_image_shapes, **params):
"""Convert predicted output tensors to final detections. Unused.
Args:
prediction_dict: a dictionary holding prediction tensors.
true_image_shapes: int32 tensor of shape [batch, 3] where each row is
of the form [height, width, channels] indicating the shapes
of true images in the resized images, as resized images can be padded
with zeros.
**params: Additional keyword arguments for specific implementations of
DetectionModel.
Returns:
detections: a dictionary with empty fields.
"""
return {
'detection_boxes': None,
'detection_scores': None,
'detection_classes': None,
'num_detections': None
}
def loss(self, prediction_dict, true_image_shapes):
"""Compute scalar loss tensors with respect to provided groundtruth.
Calling this function requires that groundtruth tensors have been
provided via the provide_groundtruth function.
Args:
prediction_dict: a dictionary holding predicted tensors
true_image_shapes: int32 tensor of shape [batch, 3] where each row is
of the form [height, width, channels] indicating the shapes
of true images in the resized images, as resized images can be padded
with zeros.
Returns:
a dictionary mapping strings (loss names) to scalar tensors representing
loss values.
"""
batch_reg_targets = tf.stack(
self.groundtruth_lists(fields.BoxListFields.boxes))
batch_cls_targets = tf.stack(
self.groundtruth_lists(fields.BoxListFields.classes))
weights = tf.constant(
1.0, dtype=tf.float32,
shape=[len(self.groundtruth_lists(fields.BoxListFields.boxes)), 1])
location_losses = self._localization_loss(
prediction_dict['box_encodings'], batch_reg_targets,
weights=weights)
cls_losses = self._classification_loss(
prediction_dict['class_predictions_with_background'], batch_cls_targets,
weights=weights)
loss_dict = {
'localization_loss': tf.reduce_sum(location_losses),
'classification_loss': tf.reduce_sum(cls_losses),
}
return loss_dict
def regularization_losses(self):
"""Returns a list of regularization losses for this model.
Returns a list of regularization losses for this model that the estimator
needs to use during training/optimization.
Returns:
A list of regularization loss tensors.
"""
pass
def restore_map(self, fine_tune_checkpoint_type='detection'):
"""Returns a map of variables to load from a foreign checkpoint.
Args:
fine_tune_checkpoint_type: whether to restore from a full detection
checkpoint (with compatible variable names) or to restore from a
classification checkpoint for initialization prior to training.
Valid values: `detection`, `classification`. Default 'detection'.
Returns:
A dict mapping variable names to variables.
"""
return {var.op.name: var for var in tf.global_variables()}
def updates(self):
"""Returns a list of update operators for this model.
Returns a list of update operators for this model that must be executed at
each training step. The estimator's train op needs to have a control
dependency on these updates.
Returns:
A list of update operators.
"""
pass
@unittest.skipIf(tf_version.is_tf2(), 'Skipping TF1.X only test.')
class TrainerTest(tf.test.TestCase):
def test_configure_trainer_and_train_two_steps(self):
train_config_text_proto = """
optimizer {
adam_optimizer {
learning_rate {
constant_learning_rate {
learning_rate: 0.01
}
}
}
}
data_augmentation_options {
random_adjust_brightness {
max_delta: 0.2
}
}
data_augmentation_options {
random_adjust_contrast {
min_delta: 0.7
max_delta: 1.1
}
}
num_steps: 2
"""
train_config = train_pb2.TrainConfig()
text_format.Merge(train_config_text_proto, train_config)
train_dir = self.get_temp_dir()
trainer.train(
create_tensor_dict_fn=get_input_function,
create_model_fn=FakeDetectionModel,
train_config=train_config,
master='',
task=0,
num_clones=1,
worker_replicas=1,
clone_on_cpu=True,
ps_tasks=0,
worker_job_name='worker',
is_chief=True,
train_dir=train_dir)
def test_configure_trainer_with_multiclass_scores_and_train_two_steps(self):
train_config_text_proto = """
optimizer {
adam_optimizer {
learning_rate {
constant_learning_rate {
learning_rate: 0.01
}
}
}
}
data_augmentation_options {
random_adjust_brightness {
max_delta: 0.2
}
}
data_augmentation_options {
random_adjust_contrast {
min_delta: 0.7
max_delta: 1.1
}
}
num_steps: 2
use_multiclass_scores: true
"""
train_config = train_pb2.TrainConfig()
text_format.Merge(train_config_text_proto, train_config)
train_dir = self.get_temp_dir()
trainer.train(create_tensor_dict_fn=get_input_function,
create_model_fn=FakeDetectionModel,
train_config=train_config,
master='',
task=0,
num_clones=1,
worker_replicas=1,
clone_on_cpu=True,
ps_tasks=0,
worker_job_name='worker',
is_chief=True,
train_dir=train_dir)
if __name__ == '__main__':
tf.test.main()
|