"llm/llama.cpp/ggml/src/ggml-cuda/tsembd.cuh" did not exist on "ecd2f176277db4f074e25a2c3646b04b51cec119"
exporter_test.py 42.7 KB
Newer Older
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
# 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.export_inference_graph."""
import os
import numpy as np
19
import six
20
import tensorflow as tf
21
from google.protobuf import text_format
22
23
24
25
26
from object_detection import exporter
from object_detection.builders import model_builder
from object_detection.core import model
from object_detection.protos import pipeline_pb2

27
28
29
30
31
if six.PY2:
  import mock  # pylint: disable=g-import-not-at-top
else:
  from unittest import mock  # pylint: disable=g-import-not-at-top

Vivek Rathod's avatar
Vivek Rathod committed
32
33
slim = tf.contrib.slim

34
35
36

class FakeModel(model.DetectionModel):

37
38
  def __init__(self, add_detection_keypoints=False, add_detection_masks=False):
    self._add_detection_keypoints = add_detection_keypoints
39
40
    self._add_detection_masks = add_detection_masks

41
  def preprocess(self, inputs):
42
43
    true_image_shapes = []  # Doesn't matter for the fake model.
    return tf.identity(inputs), true_image_shapes
44

45
  def predict(self, preprocessed_inputs, true_image_shapes):
46
    return {'image': tf.layers.conv2d(preprocessed_inputs, 3, 1)}
47

48
  def postprocess(self, prediction_dict, true_image_shapes):
49
    with tf.control_dependencies(prediction_dict.values()):
50
      postprocessed_tensors = {
51
52
53
54
55
56
57
58
59
          'detection_boxes': tf.constant([[[0.0, 0.0, 0.5, 0.5],
                                           [0.5, 0.5, 0.8, 0.8]],
                                          [[0.5, 0.5, 1.0, 1.0],
                                           [0.0, 0.0, 0.0, 0.0]]], tf.float32),
          'detection_scores': tf.constant([[0.7, 0.6],
                                           [0.9, 0.0]], tf.float32),
          'detection_classes': tf.constant([[0, 1],
                                            [1, 0]], tf.float32),
          'num_detections': tf.constant([2, 1], tf.float32)
60
      }
61
62
63
      if self._add_detection_keypoints:
        postprocessed_tensors['detection_keypoints'] = tf.constant(
            np.arange(48).reshape([2, 2, 6, 2]), tf.float32)
64
65
      if self._add_detection_masks:
        postprocessed_tensors['detection_masks'] = tf.constant(
66
            np.arange(64).reshape([2, 2, 4, 4]), tf.float32)
67
    return postprocessed_tensors
68

69
  def restore_map(self, checkpoint_path, fine_tune_checkpoint_type):
70
71
    pass

72
  def loss(self, prediction_dict, true_image_shapes):
73
74
75
76
77
78
79
80
81
    pass


class ExportInferenceGraphTest(tf.test.TestCase):

  def _save_checkpoint_from_mock_model(self, checkpoint_path,
                                       use_moving_averages):
    g = tf.Graph()
    with g.as_default():
82
      mock_model = FakeModel()
83
      preprocessed_inputs, true_image_shapes = mock_model.preprocess(
84
          tf.placeholder(tf.float32, shape=[None, None, None, 3]))
85
86
      predictions = mock_model.predict(preprocessed_inputs, true_image_shapes)
      mock_model.postprocess(predictions, true_image_shapes)
87
88
      if use_moving_averages:
        tf.train.ExponentialMovingAverage(0.0).apply()
Vivek Rathod's avatar
Vivek Rathod committed
89
      slim.get_or_create_global_step()
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
      saver = tf.train.Saver()
      init = tf.global_variables_initializer()
      with self.test_session() as sess:
        sess.run(init)
        saver.save(sess, checkpoint_path)

  def _load_inference_graph(self, inference_graph_path):
    od_graph = tf.Graph()
    with od_graph.as_default():
      od_graph_def = tf.GraphDef()
      with tf.gfile.GFile(inference_graph_path) as fid:
        serialized_graph = fid.read()
        od_graph_def.ParseFromString(serialized_graph)
        tf.import_graph_def(od_graph_def, name='')
    return od_graph

  def _create_tf_example(self, image_array):
    with self.test_session():
      encoded_image = tf.image.encode_jpeg(tf.constant(image_array)).eval()
    def _bytes_feature(value):
      return tf.train.Feature(bytes_list=tf.train.BytesList(value=[value]))
    example = tf.train.Example(features=tf.train.Features(feature={
        'image/encoded': _bytes_feature(encoded_image),
        'image/format': _bytes_feature('jpg'),
        'image/source_id': _bytes_feature('image_id')
    })).SerializeToString()
    return example

  def test_export_graph_with_image_tensor_input(self):
119
120
121
122
    tmp_dir = self.get_temp_dir()
    trained_checkpoint_prefix = os.path.join(tmp_dir, 'model.ckpt')
    self._save_checkpoint_from_mock_model(trained_checkpoint_prefix,
                                          use_moving_averages=False)
123
124
    with mock.patch.object(
        model_builder, 'build', autospec=True) as mock_builder:
125
      mock_builder.return_value = FakeModel()
126
      output_directory = os.path.join(tmp_dir, 'output')
127
128
129
130
131
      pipeline_config = pipeline_pb2.TrainEvalPipelineConfig()
      pipeline_config.eval_config.use_moving_averages = False
      exporter.export_inference_graph(
          input_type='image_tensor',
          pipeline_config=pipeline_config,
132
133
          trained_checkpoint_prefix=trained_checkpoint_prefix,
          output_directory=output_directory)
Vivek Rathod's avatar
Vivek Rathod committed
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
      self.assertTrue(os.path.exists(os.path.join(
          output_directory, 'saved_model', 'saved_model.pb')))

  def test_export_graph_with_fixed_size_image_tensor_input(self):
    input_shape = [1, 320, 320, 3]

    tmp_dir = self.get_temp_dir()
    trained_checkpoint_prefix = os.path.join(tmp_dir, 'model.ckpt')
    self._save_checkpoint_from_mock_model(
        trained_checkpoint_prefix, use_moving_averages=False)
    with mock.patch.object(
        model_builder, 'build', autospec=True) as mock_builder:
      mock_builder.return_value = FakeModel()
      output_directory = os.path.join(tmp_dir, 'output')
      pipeline_config = pipeline_pb2.TrainEvalPipelineConfig()
      pipeline_config.eval_config.use_moving_averages = False
      exporter.export_inference_graph(
          input_type='image_tensor',
          pipeline_config=pipeline_config,
          trained_checkpoint_prefix=trained_checkpoint_prefix,
          output_directory=output_directory,
          input_shape=input_shape)
      saved_model_path = os.path.join(output_directory, 'saved_model')
      self.assertTrue(
          os.path.exists(os.path.join(saved_model_path, 'saved_model.pb')))

    with tf.Graph().as_default() as od_graph:
      with self.test_session(graph=od_graph) as sess:
        meta_graph = tf.saved_model.loader.load(
            sess, [tf.saved_model.tag_constants.SERVING], saved_model_path)
        signature = meta_graph.signature_def['serving_default']
        input_tensor_name = signature.inputs['inputs'].name
        image_tensor = od_graph.get_tensor_by_name(input_tensor_name)
        self.assertSequenceEqual(image_tensor.get_shape().as_list(),
                                 input_shape)
169
170

  def test_export_graph_with_tf_example_input(self):
171
172
173
174
    tmp_dir = self.get_temp_dir()
    trained_checkpoint_prefix = os.path.join(tmp_dir, 'model.ckpt')
    self._save_checkpoint_from_mock_model(trained_checkpoint_prefix,
                                          use_moving_averages=False)
175
176
    with mock.patch.object(
        model_builder, 'build', autospec=True) as mock_builder:
177
      mock_builder.return_value = FakeModel()
178
      output_directory = os.path.join(tmp_dir, 'output')
179
180
181
182
183
      pipeline_config = pipeline_pb2.TrainEvalPipelineConfig()
      pipeline_config.eval_config.use_moving_averages = False
      exporter.export_inference_graph(
          input_type='tf_example',
          pipeline_config=pipeline_config,
184
185
          trained_checkpoint_prefix=trained_checkpoint_prefix,
          output_directory=output_directory)
Vivek Rathod's avatar
Vivek Rathod committed
186
187
      self.assertTrue(os.path.exists(os.path.join(
          output_directory, 'saved_model', 'saved_model.pb')))
188

189
  def test_export_graph_with_encoded_image_string_input(self):
190
191
192
    tmp_dir = self.get_temp_dir()
    trained_checkpoint_prefix = os.path.join(tmp_dir, 'model.ckpt')
    self._save_checkpoint_from_mock_model(trained_checkpoint_prefix,
193
194
195
                                          use_moving_averages=False)
    with mock.patch.object(
        model_builder, 'build', autospec=True) as mock_builder:
196
      mock_builder.return_value = FakeModel()
197
      output_directory = os.path.join(tmp_dir, 'output')
198
199
200
      pipeline_config = pipeline_pb2.TrainEvalPipelineConfig()
      pipeline_config.eval_config.use_moving_averages = False
      exporter.export_inference_graph(
201
          input_type='encoded_image_string_tensor',
202
          pipeline_config=pipeline_config,
203
204
          trained_checkpoint_prefix=trained_checkpoint_prefix,
          output_directory=output_directory)
Vivek Rathod's avatar
Vivek Rathod committed
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
      self.assertTrue(os.path.exists(os.path.join(
          output_directory, 'saved_model', 'saved_model.pb')))

  def _get_variables_in_checkpoint(self, checkpoint_file):
    return set([
        var_name
        for var_name, _ in tf.train.list_variables(checkpoint_file)])

  def test_replace_variable_values_with_moving_averages(self):
    tmp_dir = self.get_temp_dir()
    trained_checkpoint_prefix = os.path.join(tmp_dir, 'model.ckpt')
    new_checkpoint_prefix = os.path.join(tmp_dir, 'new.ckpt')
    self._save_checkpoint_from_mock_model(trained_checkpoint_prefix,
                                          use_moving_averages=True)
    graph = tf.Graph()
    with graph.as_default():
      fake_model = FakeModel()
222
      preprocessed_inputs, true_image_shapes = fake_model.preprocess(
Vivek Rathod's avatar
Vivek Rathod committed
223
          tf.placeholder(dtype=tf.float32, shape=[None, None, None, 3]))
224
225
      predictions = fake_model.predict(preprocessed_inputs, true_image_shapes)
      fake_model.postprocess(predictions, true_image_shapes)
Vivek Rathod's avatar
Vivek Rathod committed
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
      exporter.replace_variable_values_with_moving_averages(
          graph, trained_checkpoint_prefix, new_checkpoint_prefix)

    expected_variables = set(['conv2d/bias', 'conv2d/kernel'])
    variables_in_old_ckpt = self._get_variables_in_checkpoint(
        trained_checkpoint_prefix)
    self.assertIn('conv2d/bias/ExponentialMovingAverage',
                  variables_in_old_ckpt)
    self.assertIn('conv2d/kernel/ExponentialMovingAverage',
                  variables_in_old_ckpt)
    variables_in_new_ckpt = self._get_variables_in_checkpoint(
        new_checkpoint_prefix)
    self.assertTrue(expected_variables.issubset(variables_in_new_ckpt))
    self.assertNotIn('conv2d/bias/ExponentialMovingAverage',
                     variables_in_new_ckpt)
    self.assertNotIn('conv2d/kernel/ExponentialMovingAverage',
                     variables_in_new_ckpt)
243

244
245
246
247
  def test_export_graph_with_moving_averages(self):
    tmp_dir = self.get_temp_dir()
    trained_checkpoint_prefix = os.path.join(tmp_dir, 'model.ckpt')
    self._save_checkpoint_from_mock_model(trained_checkpoint_prefix,
248
                                          use_moving_averages=True)
249
    output_directory = os.path.join(tmp_dir, 'output')
250
251
    with mock.patch.object(
        model_builder, 'build', autospec=True) as mock_builder:
252
      mock_builder.return_value = FakeModel()
253
254
255
256
257
      pipeline_config = pipeline_pb2.TrainEvalPipelineConfig()
      pipeline_config.eval_config.use_moving_averages = True
      exporter.export_inference_graph(
          input_type='image_tensor',
          pipeline_config=pipeline_config,
258
259
          trained_checkpoint_prefix=trained_checkpoint_prefix,
          output_directory=output_directory)
Vivek Rathod's avatar
Vivek Rathod committed
260
261
262
263
264
265
      self.assertTrue(os.path.exists(os.path.join(
          output_directory, 'saved_model', 'saved_model.pb')))
    expected_variables = set(['conv2d/bias', 'conv2d/kernel', 'global_step'])
    actual_variables = set(
        [var_name for var_name, _ in tf.train.list_variables(output_directory)])
    self.assertTrue(expected_variables.issubset(actual_variables))
266

267
  def test_export_model_with_all_output_nodes(self):
268
269
270
271
272
273
274
    tmp_dir = self.get_temp_dir()
    trained_checkpoint_prefix = os.path.join(tmp_dir, 'model.ckpt')
    self._save_checkpoint_from_mock_model(trained_checkpoint_prefix,
                                          use_moving_averages=True)
    output_directory = os.path.join(tmp_dir, 'output')
    inference_graph_path = os.path.join(output_directory,
                                        'frozen_inference_graph.pb')
275
276
    with mock.patch.object(
        model_builder, 'build', autospec=True) as mock_builder:
277
278
      mock_builder.return_value = FakeModel(
          add_detection_keypoints=True, add_detection_masks=True)
279
280
281
282
      pipeline_config = pipeline_pb2.TrainEvalPipelineConfig()
      exporter.export_inference_graph(
          input_type='image_tensor',
          pipeline_config=pipeline_config,
283
284
          trained_checkpoint_prefix=trained_checkpoint_prefix,
          output_directory=output_directory)
285
286
287
288
289
290
    inference_graph = self._load_inference_graph(inference_graph_path)
    with self.test_session(graph=inference_graph):
      inference_graph.get_tensor_by_name('image_tensor:0')
      inference_graph.get_tensor_by_name('detection_boxes:0')
      inference_graph.get_tensor_by_name('detection_scores:0')
      inference_graph.get_tensor_by_name('detection_classes:0')
291
      inference_graph.get_tensor_by_name('detection_keypoints:0')
292
293
294
295
      inference_graph.get_tensor_by_name('detection_masks:0')
      inference_graph.get_tensor_by_name('num_detections:0')

  def test_export_model_with_detection_only_nodes(self):
296
297
298
299
300
301
302
    tmp_dir = self.get_temp_dir()
    trained_checkpoint_prefix = os.path.join(tmp_dir, 'model.ckpt')
    self._save_checkpoint_from_mock_model(trained_checkpoint_prefix,
                                          use_moving_averages=True)
    output_directory = os.path.join(tmp_dir, 'output')
    inference_graph_path = os.path.join(output_directory,
                                        'frozen_inference_graph.pb')
303
304
305
306
307
308
309
    with mock.patch.object(
        model_builder, 'build', autospec=True) as mock_builder:
      mock_builder.return_value = FakeModel(add_detection_masks=False)
      pipeline_config = pipeline_pb2.TrainEvalPipelineConfig()
      exporter.export_inference_graph(
          input_type='image_tensor',
          pipeline_config=pipeline_config,
310
311
          trained_checkpoint_prefix=trained_checkpoint_prefix,
          output_directory=output_directory)
312
313
314
315
316
317
318
319
    inference_graph = self._load_inference_graph(inference_graph_path)
    with self.test_session(graph=inference_graph):
      inference_graph.get_tensor_by_name('image_tensor:0')
      inference_graph.get_tensor_by_name('detection_boxes:0')
      inference_graph.get_tensor_by_name('detection_scores:0')
      inference_graph.get_tensor_by_name('detection_classes:0')
      inference_graph.get_tensor_by_name('num_detections:0')
      with self.assertRaises(KeyError):
320
        inference_graph.get_tensor_by_name('detection_keypoints:0')
321
322
        inference_graph.get_tensor_by_name('detection_masks:0')

323
  def test_export_and_run_inference_with_image_tensor(self):
324
325
326
327
328
329
330
    tmp_dir = self.get_temp_dir()
    trained_checkpoint_prefix = os.path.join(tmp_dir, 'model.ckpt')
    self._save_checkpoint_from_mock_model(trained_checkpoint_prefix,
                                          use_moving_averages=True)
    output_directory = os.path.join(tmp_dir, 'output')
    inference_graph_path = os.path.join(output_directory,
                                        'frozen_inference_graph.pb')
331
332
    with mock.patch.object(
        model_builder, 'build', autospec=True) as mock_builder:
333
334
      mock_builder.return_value = FakeModel(
          add_detection_keypoints=True, add_detection_masks=True)
335
336
337
338
339
      pipeline_config = pipeline_pb2.TrainEvalPipelineConfig()
      pipeline_config.eval_config.use_moving_averages = False
      exporter.export_inference_graph(
          input_type='image_tensor',
          pipeline_config=pipeline_config,
340
341
          trained_checkpoint_prefix=trained_checkpoint_prefix,
          output_directory=output_directory)
342
343
344
345
346
347
348

    inference_graph = self._load_inference_graph(inference_graph_path)
    with self.test_session(graph=inference_graph) as sess:
      image_tensor = inference_graph.get_tensor_by_name('image_tensor:0')
      boxes = inference_graph.get_tensor_by_name('detection_boxes:0')
      scores = inference_graph.get_tensor_by_name('detection_scores:0')
      classes = inference_graph.get_tensor_by_name('detection_classes:0')
349
      keypoints = inference_graph.get_tensor_by_name('detection_keypoints:0')
350
      masks = inference_graph.get_tensor_by_name('detection_masks:0')
351
      num_detections = inference_graph.get_tensor_by_name('num_detections:0')
352
353
354
355
      (boxes_np, scores_np, classes_np, keypoints_np, masks_np,
       num_detections_np) = sess.run(
           [boxes, scores, classes, keypoints, masks, num_detections],
           feed_dict={image_tensor: np.ones((2, 4, 4, 3)).astype(np.uint8)})
356
357
358
359
360
361
362
363
      self.assertAllClose(boxes_np, [[[0.0, 0.0, 0.5, 0.5],
                                      [0.5, 0.5, 0.8, 0.8]],
                                     [[0.5, 0.5, 1.0, 1.0],
                                      [0.0, 0.0, 0.0, 0.0]]])
      self.assertAllClose(scores_np, [[0.7, 0.6],
                                      [0.9, 0.0]])
      self.assertAllClose(classes_np, [[1, 2],
                                       [2, 1]])
364
      self.assertAllClose(keypoints_np, np.arange(48).reshape([2, 2, 6, 2]))
365
366
      self.assertAllClose(masks_np, np.arange(64).reshape([2, 2, 4, 4]))
      self.assertAllClose(num_detections_np, [2, 1])
367

368
369
370
371
372
373
374
375
376
377
378
379
380
  def _create_encoded_image_string(self, image_array_np, encoding_format):
    od_graph = tf.Graph()
    with od_graph.as_default():
      if encoding_format == 'jpg':
        encoded_string = tf.image.encode_jpeg(image_array_np)
      elif encoding_format == 'png':
        encoded_string = tf.image.encode_png(image_array_np)
      else:
        raise ValueError('Supports only the following formats: `jpg`, `png`')
    with self.test_session(graph=od_graph):
      return encoded_string.eval()

  def test_export_and_run_inference_with_encoded_image_string_tensor(self):
381
382
383
384
385
386
387
    tmp_dir = self.get_temp_dir()
    trained_checkpoint_prefix = os.path.join(tmp_dir, 'model.ckpt')
    self._save_checkpoint_from_mock_model(trained_checkpoint_prefix,
                                          use_moving_averages=True)
    output_directory = os.path.join(tmp_dir, 'output')
    inference_graph_path = os.path.join(output_directory,
                                        'frozen_inference_graph.pb')
388
389
    with mock.patch.object(
        model_builder, 'build', autospec=True) as mock_builder:
390
391
      mock_builder.return_value = FakeModel(
          add_detection_keypoints=True, add_detection_masks=True)
392
393
394
395
396
      pipeline_config = pipeline_pb2.TrainEvalPipelineConfig()
      pipeline_config.eval_config.use_moving_averages = False
      exporter.export_inference_graph(
          input_type='encoded_image_string_tensor',
          pipeline_config=pipeline_config,
397
398
          trained_checkpoint_prefix=trained_checkpoint_prefix,
          output_directory=output_directory)
399
400
401
402
403
404
405
406
407
408
409
410

    inference_graph = self._load_inference_graph(inference_graph_path)
    jpg_image_str = self._create_encoded_image_string(
        np.ones((4, 4, 3)).astype(np.uint8), 'jpg')
    png_image_str = self._create_encoded_image_string(
        np.ones((4, 4, 3)).astype(np.uint8), 'png')
    with self.test_session(graph=inference_graph) as sess:
      image_str_tensor = inference_graph.get_tensor_by_name(
          'encoded_image_string_tensor:0')
      boxes = inference_graph.get_tensor_by_name('detection_boxes:0')
      scores = inference_graph.get_tensor_by_name('detection_scores:0')
      classes = inference_graph.get_tensor_by_name('detection_classes:0')
411
      keypoints = inference_graph.get_tensor_by_name('detection_keypoints:0')
412
413
414
      masks = inference_graph.get_tensor_by_name('detection_masks:0')
      num_detections = inference_graph.get_tensor_by_name('num_detections:0')
      for image_str in [jpg_image_str, png_image_str]:
415
        image_str_batch_np = np.hstack([image_str]* 2)
416
        (boxes_np, scores_np, classes_np, keypoints_np, masks_np,
417
         num_detections_np) = sess.run(
418
             [boxes, scores, classes, keypoints, masks, num_detections],
419
420
421
422
423
424
425
426
427
             feed_dict={image_str_tensor: image_str_batch_np})
        self.assertAllClose(boxes_np, [[[0.0, 0.0, 0.5, 0.5],
                                        [0.5, 0.5, 0.8, 0.8]],
                                       [[0.5, 0.5, 1.0, 1.0],
                                        [0.0, 0.0, 0.0, 0.0]]])
        self.assertAllClose(scores_np, [[0.7, 0.6],
                                        [0.9, 0.0]])
        self.assertAllClose(classes_np, [[1, 2],
                                         [2, 1]])
428
        self.assertAllClose(keypoints_np, np.arange(48).reshape([2, 2, 6, 2]))
429
430
431
432
433
434
435
436
437
438
439
440
441
        self.assertAllClose(masks_np, np.arange(64).reshape([2, 2, 4, 4]))
        self.assertAllClose(num_detections_np, [2, 1])

  def test_raise_runtime_error_on_images_with_different_sizes(self):
    tmp_dir = self.get_temp_dir()
    trained_checkpoint_prefix = os.path.join(tmp_dir, 'model.ckpt')
    self._save_checkpoint_from_mock_model(trained_checkpoint_prefix,
                                          use_moving_averages=True)
    output_directory = os.path.join(tmp_dir, 'output')
    inference_graph_path = os.path.join(output_directory,
                                        'frozen_inference_graph.pb')
    with mock.patch.object(
        model_builder, 'build', autospec=True) as mock_builder:
442
443
      mock_builder.return_value = FakeModel(
          add_detection_keypoints=True, add_detection_masks=True)
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
      pipeline_config = pipeline_pb2.TrainEvalPipelineConfig()
      pipeline_config.eval_config.use_moving_averages = False
      exporter.export_inference_graph(
          input_type='encoded_image_string_tensor',
          pipeline_config=pipeline_config,
          trained_checkpoint_prefix=trained_checkpoint_prefix,
          output_directory=output_directory)

    inference_graph = self._load_inference_graph(inference_graph_path)
    large_image = self._create_encoded_image_string(
        np.ones((4, 4, 3)).astype(np.uint8), 'jpg')
    small_image = self._create_encoded_image_string(
        np.ones((2, 2, 3)).astype(np.uint8), 'jpg')

    image_str_batch_np = np.hstack([large_image, small_image])
    with self.test_session(graph=inference_graph) as sess:
      image_str_tensor = inference_graph.get_tensor_by_name(
          'encoded_image_string_tensor:0')
      boxes = inference_graph.get_tensor_by_name('detection_boxes:0')
      scores = inference_graph.get_tensor_by_name('detection_scores:0')
      classes = inference_graph.get_tensor_by_name('detection_classes:0')
465
      keypoints = inference_graph.get_tensor_by_name('detection_keypoints:0')
466
467
468
      masks = inference_graph.get_tensor_by_name('detection_masks:0')
      num_detections = inference_graph.get_tensor_by_name('num_detections:0')
      with self.assertRaisesRegexp(tf.errors.InvalidArgumentError,
469
                                   'TensorArray.*shape'):
470
471
472
        sess.run(
            [boxes, scores, classes, keypoints, masks, num_detections],
            feed_dict={image_str_tensor: image_str_batch_np})
473

474
  def test_export_and_run_inference_with_tf_example(self):
475
476
477
478
479
480
481
    tmp_dir = self.get_temp_dir()
    trained_checkpoint_prefix = os.path.join(tmp_dir, 'model.ckpt')
    self._save_checkpoint_from_mock_model(trained_checkpoint_prefix,
                                          use_moving_averages=True)
    output_directory = os.path.join(tmp_dir, 'output')
    inference_graph_path = os.path.join(output_directory,
                                        'frozen_inference_graph.pb')
482
483
    with mock.patch.object(
        model_builder, 'build', autospec=True) as mock_builder:
484
485
      mock_builder.return_value = FakeModel(
          add_detection_keypoints=True, add_detection_masks=True)
486
487
488
489
490
      pipeline_config = pipeline_pb2.TrainEvalPipelineConfig()
      pipeline_config.eval_config.use_moving_averages = False
      exporter.export_inference_graph(
          input_type='tf_example',
          pipeline_config=pipeline_config,
491
492
          trained_checkpoint_prefix=trained_checkpoint_prefix,
          output_directory=output_directory)
493
494

    inference_graph = self._load_inference_graph(inference_graph_path)
495
496
    tf_example_np = np.expand_dims(self._create_tf_example(
        np.ones((4, 4, 3)).astype(np.uint8)), axis=0)
497
498
499
500
501
    with self.test_session(graph=inference_graph) as sess:
      tf_example = inference_graph.get_tensor_by_name('tf_example:0')
      boxes = inference_graph.get_tensor_by_name('detection_boxes:0')
      scores = inference_graph.get_tensor_by_name('detection_scores:0')
      classes = inference_graph.get_tensor_by_name('detection_classes:0')
502
      keypoints = inference_graph.get_tensor_by_name('detection_keypoints:0')
503
      masks = inference_graph.get_tensor_by_name('detection_masks:0')
504
      num_detections = inference_graph.get_tensor_by_name('num_detections:0')
505
506
507
508
      (boxes_np, scores_np, classes_np, keypoints_np, masks_np,
       num_detections_np) = sess.run(
           [boxes, scores, classes, keypoints, masks, num_detections],
           feed_dict={tf_example: tf_example_np})
509
510
511
512
513
514
515
516
      self.assertAllClose(boxes_np, [[[0.0, 0.0, 0.5, 0.5],
                                      [0.5, 0.5, 0.8, 0.8]],
                                     [[0.5, 0.5, 1.0, 1.0],
                                      [0.0, 0.0, 0.0, 0.0]]])
      self.assertAllClose(scores_np, [[0.7, 0.6],
                                      [0.9, 0.0]])
      self.assertAllClose(classes_np, [[1, 2],
                                       [2, 1]])
517
      self.assertAllClose(keypoints_np, np.arange(48).reshape([2, 2, 6, 2]))
518
519
      self.assertAllClose(masks_np, np.arange(64).reshape([2, 2, 4, 4]))
      self.assertAllClose(num_detections_np, [2, 1])
520

521
522
523
524
525
526
527
528
529
530
531
  def test_write_frozen_graph(self):
    tmp_dir = self.get_temp_dir()
    trained_checkpoint_prefix = os.path.join(tmp_dir, 'model.ckpt')
    self._save_checkpoint_from_mock_model(trained_checkpoint_prefix,
                                          use_moving_averages=True)
    output_directory = os.path.join(tmp_dir, 'output')
    inference_graph_path = os.path.join(output_directory,
                                        'frozen_inference_graph.pb')
    tf.gfile.MakeDirs(output_directory)
    with mock.patch.object(
        model_builder, 'build', autospec=True) as mock_builder:
532
533
      mock_builder.return_value = FakeModel(
          add_detection_keypoints=True, add_detection_masks=True)
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
      pipeline_config = pipeline_pb2.TrainEvalPipelineConfig()
      pipeline_config.eval_config.use_moving_averages = False
      detection_model = model_builder.build(pipeline_config.model,
                                            is_training=False)
      outputs, _ = exporter._build_detection_graph(
          input_type='tf_example',
          detection_model=detection_model,
          input_shape=None,
          output_collection_name='inference_op',
          graph_hook_fn=None)
      output_node_names = ','.join(outputs.keys())
      saver = tf.train.Saver()
      input_saver_def = saver.as_saver_def()
      frozen_graph_def = exporter.freeze_graph_with_def_protos(
          input_graph_def=tf.get_default_graph().as_graph_def(),
          input_saver_def=input_saver_def,
          input_checkpoint=trained_checkpoint_prefix,
          output_node_names=output_node_names,
          restore_op_name='save/restore_all',
          filename_tensor_name='save/Const:0',
          clear_devices=True,
          initializer_nodes='')
      exporter.write_frozen_graph(inference_graph_path, frozen_graph_def)

    inference_graph = self._load_inference_graph(inference_graph_path)
    tf_example_np = np.expand_dims(self._create_tf_example(
        np.ones((4, 4, 3)).astype(np.uint8)), axis=0)
    with self.test_session(graph=inference_graph) as sess:
      tf_example = inference_graph.get_tensor_by_name('tf_example:0')
      boxes = inference_graph.get_tensor_by_name('detection_boxes:0')
      scores = inference_graph.get_tensor_by_name('detection_scores:0')
      classes = inference_graph.get_tensor_by_name('detection_classes:0')
566
      keypoints = inference_graph.get_tensor_by_name('detection_keypoints:0')
567
568
      masks = inference_graph.get_tensor_by_name('detection_masks:0')
      num_detections = inference_graph.get_tensor_by_name('num_detections:0')
569
570
571
572
      (boxes_np, scores_np, classes_np, keypoints_np, masks_np,
       num_detections_np) = sess.run(
           [boxes, scores, classes, keypoints, masks, num_detections],
           feed_dict={tf_example: tf_example_np})
573
574
575
576
577
578
579
580
      self.assertAllClose(boxes_np, [[[0.0, 0.0, 0.5, 0.5],
                                      [0.5, 0.5, 0.8, 0.8]],
                                     [[0.5, 0.5, 1.0, 1.0],
                                      [0.0, 0.0, 0.0, 0.0]]])
      self.assertAllClose(scores_np, [[0.7, 0.6],
                                      [0.9, 0.0]])
      self.assertAllClose(classes_np, [[1, 2],
                                       [2, 1]])
581
      self.assertAllClose(keypoints_np, np.arange(48).reshape([2, 2, 6, 2]))
582
583
584
      self.assertAllClose(masks_np, np.arange(64).reshape([2, 2, 4, 4]))
      self.assertAllClose(num_detections_np, [2, 1])

585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
  def test_export_graph_saves_pipeline_file(self):
    tmp_dir = self.get_temp_dir()
    trained_checkpoint_prefix = os.path.join(tmp_dir, 'model.ckpt')
    self._save_checkpoint_from_mock_model(trained_checkpoint_prefix,
                                          use_moving_averages=True)
    output_directory = os.path.join(tmp_dir, 'output')
    with mock.patch.object(
        model_builder, 'build', autospec=True) as mock_builder:
      mock_builder.return_value = FakeModel()
      pipeline_config = pipeline_pb2.TrainEvalPipelineConfig()
      exporter.export_inference_graph(
          input_type='image_tensor',
          pipeline_config=pipeline_config,
          trained_checkpoint_prefix=trained_checkpoint_prefix,
          output_directory=output_directory)
      expected_pipeline_path = os.path.join(
          output_directory, 'pipeline.config')
      self.assertTrue(os.path.exists(expected_pipeline_path))

      written_pipeline_config = pipeline_pb2.TrainEvalPipelineConfig()
      with tf.gfile.GFile(expected_pipeline_path, 'r') as f:
        proto_str = f.read()
        text_format.Merge(proto_str, written_pipeline_config)
        self.assertProtoEquals(pipeline_config, written_pipeline_config)

610
  def test_export_saved_model_and_run_inference(self):
611
612
613
    tmp_dir = self.get_temp_dir()
    trained_checkpoint_prefix = os.path.join(tmp_dir, 'model.ckpt')
    self._save_checkpoint_from_mock_model(trained_checkpoint_prefix,
614
                                          use_moving_averages=False)
615
616
    output_directory = os.path.join(tmp_dir, 'output')
    saved_model_path = os.path.join(output_directory, 'saved_model')
617
618
619

    with mock.patch.object(
        model_builder, 'build', autospec=True) as mock_builder:
620
621
      mock_builder.return_value = FakeModel(
          add_detection_keypoints=True, add_detection_masks=True)
622
623
624
625
626
      pipeline_config = pipeline_pb2.TrainEvalPipelineConfig()
      pipeline_config.eval_config.use_moving_averages = False
      exporter.export_inference_graph(
          input_type='tf_example',
          pipeline_config=pipeline_config,
627
628
          trained_checkpoint_prefix=trained_checkpoint_prefix,
          output_directory=output_directory)
629

630
631
    tf_example_np = np.hstack([self._create_tf_example(
        np.ones((4, 4, 3)).astype(np.uint8))] * 2)
632
633
    with tf.Graph().as_default() as od_graph:
      with self.test_session(graph=od_graph) as sess:
Vivek Rathod's avatar
Vivek Rathod committed
634
        meta_graph = tf.saved_model.loader.load(
635
            sess, [tf.saved_model.tag_constants.SERVING], saved_model_path)
Vivek Rathod's avatar
Vivek Rathod committed
636
637
638
639
640
641
642
643
644
645
646

        signature = meta_graph.signature_def['serving_default']
        input_tensor_name = signature.inputs['inputs'].name
        tf_example = od_graph.get_tensor_by_name(input_tensor_name)

        boxes = od_graph.get_tensor_by_name(
            signature.outputs['detection_boxes'].name)
        scores = od_graph.get_tensor_by_name(
            signature.outputs['detection_scores'].name)
        classes = od_graph.get_tensor_by_name(
            signature.outputs['detection_classes'].name)
647
648
        keypoints = od_graph.get_tensor_by_name(
            signature.outputs['detection_keypoints'].name)
Vivek Rathod's avatar
Vivek Rathod committed
649
650
651
652
653
        masks = od_graph.get_tensor_by_name(
            signature.outputs['detection_masks'].name)
        num_detections = od_graph.get_tensor_by_name(
            signature.outputs['num_detections'].name)

654
        (boxes_np, scores_np, classes_np, keypoints_np, masks_np,
655
         num_detections_np) = sess.run(
656
             [boxes, scores, classes, keypoints, masks, num_detections],
657
658
659
660
661
662
663
664
665
             feed_dict={tf_example: tf_example_np})
        self.assertAllClose(boxes_np, [[[0.0, 0.0, 0.5, 0.5],
                                        [0.5, 0.5, 0.8, 0.8]],
                                       [[0.5, 0.5, 1.0, 1.0],
                                        [0.0, 0.0, 0.0, 0.0]]])
        self.assertAllClose(scores_np, [[0.7, 0.6],
                                        [0.9, 0.0]])
        self.assertAllClose(classes_np, [[1, 2],
                                         [2, 1]])
666
        self.assertAllClose(keypoints_np, np.arange(48).reshape([2, 2, 6, 2]))
667
668
        self.assertAllClose(masks_np, np.arange(64).reshape([2, 2, 4, 4]))
        self.assertAllClose(num_detections_np, [2, 1])
669

670
671
672
673
674
675
676
677
678
679
  def test_write_saved_model(self):
    tmp_dir = self.get_temp_dir()
    trained_checkpoint_prefix = os.path.join(tmp_dir, 'model.ckpt')
    self._save_checkpoint_from_mock_model(trained_checkpoint_prefix,
                                          use_moving_averages=False)
    output_directory = os.path.join(tmp_dir, 'output')
    saved_model_path = os.path.join(output_directory, 'saved_model')
    tf.gfile.MakeDirs(output_directory)
    with mock.patch.object(
        model_builder, 'build', autospec=True) as mock_builder:
680
681
      mock_builder.return_value = FakeModel(
          add_detection_keypoints=True, add_detection_masks=True)
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
      pipeline_config = pipeline_pb2.TrainEvalPipelineConfig()
      pipeline_config.eval_config.use_moving_averages = False
      detection_model = model_builder.build(pipeline_config.model,
                                            is_training=False)
      outputs, placeholder_tensor = exporter._build_detection_graph(
          input_type='tf_example',
          detection_model=detection_model,
          input_shape=None,
          output_collection_name='inference_op',
          graph_hook_fn=None)
      output_node_names = ','.join(outputs.keys())
      saver = tf.train.Saver()
      input_saver_def = saver.as_saver_def()
      frozen_graph_def = exporter.freeze_graph_with_def_protos(
          input_graph_def=tf.get_default_graph().as_graph_def(),
          input_saver_def=input_saver_def,
          input_checkpoint=trained_checkpoint_prefix,
          output_node_names=output_node_names,
          restore_op_name='save/restore_all',
          filename_tensor_name='save/Const:0',
          clear_devices=True,
          initializer_nodes='')
      exporter.write_saved_model(
          saved_model_path=saved_model_path,
          frozen_graph_def=frozen_graph_def,
          inputs=placeholder_tensor,
          outputs=outputs)

    tf_example_np = np.hstack([self._create_tf_example(
        np.ones((4, 4, 3)).astype(np.uint8))] * 2)
    with tf.Graph().as_default() as od_graph:
      with self.test_session(graph=od_graph) as sess:
        meta_graph = tf.saved_model.loader.load(
            sess, [tf.saved_model.tag_constants.SERVING], saved_model_path)

        signature = meta_graph.signature_def['serving_default']
        input_tensor_name = signature.inputs['inputs'].name
        tf_example = od_graph.get_tensor_by_name(input_tensor_name)

        boxes = od_graph.get_tensor_by_name(
            signature.outputs['detection_boxes'].name)
        scores = od_graph.get_tensor_by_name(
            signature.outputs['detection_scores'].name)
        classes = od_graph.get_tensor_by_name(
            signature.outputs['detection_classes'].name)
727
728
        keypoints = od_graph.get_tensor_by_name(
            signature.outputs['detection_keypoints'].name)
729
730
731
732
733
        masks = od_graph.get_tensor_by_name(
            signature.outputs['detection_masks'].name)
        num_detections = od_graph.get_tensor_by_name(
            signature.outputs['num_detections'].name)

734
        (boxes_np, scores_np, classes_np, keypoints_np, masks_np,
735
         num_detections_np) = sess.run(
736
             [boxes, scores, classes, keypoints, masks, num_detections],
737
738
739
740
741
742
743
744
745
             feed_dict={tf_example: tf_example_np})
        self.assertAllClose(boxes_np, [[[0.0, 0.0, 0.5, 0.5],
                                        [0.5, 0.5, 0.8, 0.8]],
                                       [[0.5, 0.5, 1.0, 1.0],
                                        [0.0, 0.0, 0.0, 0.0]]])
        self.assertAllClose(scores_np, [[0.7, 0.6],
                                        [0.9, 0.0]])
        self.assertAllClose(classes_np, [[1, 2],
                                         [2, 1]])
746
        self.assertAllClose(keypoints_np, np.arange(48).reshape([2, 2, 6, 2]))
747
748
749
        self.assertAllClose(masks_np, np.arange(64).reshape([2, 2, 4, 4]))
        self.assertAllClose(num_detections_np, [2, 1])

750
751
752
753
754
755
756
757
758
759
760
  def test_export_checkpoint_and_run_inference(self):
    tmp_dir = self.get_temp_dir()
    trained_checkpoint_prefix = os.path.join(tmp_dir, 'model.ckpt')
    self._save_checkpoint_from_mock_model(trained_checkpoint_prefix,
                                          use_moving_averages=False)
    output_directory = os.path.join(tmp_dir, 'output')
    model_path = os.path.join(output_directory, 'model.ckpt')
    meta_graph_path = model_path + '.meta'

    with mock.patch.object(
        model_builder, 'build', autospec=True) as mock_builder:
761
762
      mock_builder.return_value = FakeModel(
          add_detection_keypoints=True, add_detection_masks=True)
763
764
765
766
767
768
769
770
      pipeline_config = pipeline_pb2.TrainEvalPipelineConfig()
      pipeline_config.eval_config.use_moving_averages = False
      exporter.export_inference_graph(
          input_type='tf_example',
          pipeline_config=pipeline_config,
          trained_checkpoint_prefix=trained_checkpoint_prefix,
          output_directory=output_directory)

771
772
    tf_example_np = np.hstack([self._create_tf_example(
        np.ones((4, 4, 3)).astype(np.uint8))] * 2)
773
774
775
776
777
778
779
780
781
    with tf.Graph().as_default() as od_graph:
      with self.test_session(graph=od_graph) as sess:
        new_saver = tf.train.import_meta_graph(meta_graph_path)
        new_saver.restore(sess, model_path)

        tf_example = od_graph.get_tensor_by_name('tf_example:0')
        boxes = od_graph.get_tensor_by_name('detection_boxes:0')
        scores = od_graph.get_tensor_by_name('detection_scores:0')
        classes = od_graph.get_tensor_by_name('detection_classes:0')
782
        keypoints = od_graph.get_tensor_by_name('detection_keypoints:0')
783
784
        masks = od_graph.get_tensor_by_name('detection_masks:0')
        num_detections = od_graph.get_tensor_by_name('num_detections:0')
785
        (boxes_np, scores_np, classes_np, keypoints_np, masks_np,
786
         num_detections_np) = sess.run(
787
             [boxes, scores, classes, keypoints, masks, num_detections],
788
789
790
791
792
793
794
795
796
             feed_dict={tf_example: tf_example_np})
        self.assertAllClose(boxes_np, [[[0.0, 0.0, 0.5, 0.5],
                                        [0.5, 0.5, 0.8, 0.8]],
                                       [[0.5, 0.5, 1.0, 1.0],
                                        [0.0, 0.0, 0.0, 0.0]]])
        self.assertAllClose(scores_np, [[0.7, 0.6],
                                        [0.9, 0.0]])
        self.assertAllClose(classes_np, [[1, 2],
                                         [2, 1]])
797
        self.assertAllClose(keypoints_np, np.arange(48).reshape([2, 2, 6, 2]))
798
799
800
801
802
803
804
805
806
807
808
809
810
811
        self.assertAllClose(masks_np, np.arange(64).reshape([2, 2, 4, 4]))
        self.assertAllClose(num_detections_np, [2, 1])

  def test_write_graph_and_checkpoint(self):
    tmp_dir = self.get_temp_dir()
    trained_checkpoint_prefix = os.path.join(tmp_dir, 'model.ckpt')
    self._save_checkpoint_from_mock_model(trained_checkpoint_prefix,
                                          use_moving_averages=False)
    output_directory = os.path.join(tmp_dir, 'output')
    model_path = os.path.join(output_directory, 'model.ckpt')
    meta_graph_path = model_path + '.meta'
    tf.gfile.MakeDirs(output_directory)
    with mock.patch.object(
        model_builder, 'build', autospec=True) as mock_builder:
812
813
      mock_builder.return_value = FakeModel(
          add_detection_keypoints=True, add_detection_masks=True)
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
      pipeline_config = pipeline_pb2.TrainEvalPipelineConfig()
      pipeline_config.eval_config.use_moving_averages = False
      detection_model = model_builder.build(pipeline_config.model,
                                            is_training=False)
      exporter._build_detection_graph(
          input_type='tf_example',
          detection_model=detection_model,
          input_shape=None,
          output_collection_name='inference_op',
          graph_hook_fn=None)
      saver = tf.train.Saver()
      input_saver_def = saver.as_saver_def()
      exporter.write_graph_and_checkpoint(
          inference_graph_def=tf.get_default_graph().as_graph_def(),
          model_path=model_path,
          input_saver_def=input_saver_def,
          trained_checkpoint_prefix=trained_checkpoint_prefix)

    tf_example_np = np.hstack([self._create_tf_example(
        np.ones((4, 4, 3)).astype(np.uint8))] * 2)
    with tf.Graph().as_default() as od_graph:
      with self.test_session(graph=od_graph) as sess:
        new_saver = tf.train.import_meta_graph(meta_graph_path)
        new_saver.restore(sess, model_path)

        tf_example = od_graph.get_tensor_by_name('tf_example:0')
        boxes = od_graph.get_tensor_by_name('detection_boxes:0')
        scores = od_graph.get_tensor_by_name('detection_scores:0')
        classes = od_graph.get_tensor_by_name('detection_classes:0')
843
        keypoints = od_graph.get_tensor_by_name('detection_keypoints:0')
844
845
        masks = od_graph.get_tensor_by_name('detection_masks:0')
        num_detections = od_graph.get_tensor_by_name('num_detections:0')
846
        (boxes_np, scores_np, classes_np, keypoints_np, masks_np,
847
         num_detections_np) = sess.run(
848
             [boxes, scores, classes, keypoints, masks, num_detections],
849
850
851
852
853
854
855
856
857
             feed_dict={tf_example: tf_example_np})
        self.assertAllClose(boxes_np, [[[0.0, 0.0, 0.5, 0.5],
                                        [0.5, 0.5, 0.8, 0.8]],
                                       [[0.5, 0.5, 1.0, 1.0],
                                        [0.0, 0.0, 0.0, 0.0]]])
        self.assertAllClose(scores_np, [[0.7, 0.6],
                                        [0.9, 0.0]])
        self.assertAllClose(classes_np, [[1, 2],
                                         [2, 1]])
858
        self.assertAllClose(keypoints_np, np.arange(48).reshape([2, 2, 6, 2]))
859
860
        self.assertAllClose(masks_np, np.arange(64).reshape([2, 2, 4, 4]))
        self.assertAllClose(num_detections_np, [2, 1])
861
862


863
864
if __name__ == '__main__':
  tf.test.main()