def test_construct_default_conv_box_predictor(self):
box_predictor_text_proto = """
convolutional_box_predictor {
conv_hyperparams {
regularizer {
l1_regularizer {
}
}
initializer {
truncated_normal_initializer {
}
}
}
}"""
box_predictor_proto = box_predictor_pb2.BoxPredictor()
text_format.Merge(box_predictor_text_proto, box_predictor_proto)
box_predictor = box_predictor_builder.build(
argscope_fn=hyperparams_builder.build,
box_predictor_config=box_predictor_proto,
is_training=True,
num_classes=90)
self.assertEqual(box_predictor._min_depth, 0)
self.assertEqual(box_predictor._max_depth, 0)
self.assertEqual(box_predictor._num_layers_before_predictor, 0)
self.assertTrue(box_predictor._use_dropout)
self.assertAlmostEqual(box_predictor._dropout_keep_prob, 0.8)
self.assertFalse(box_predictor._apply_sigmoid_to_scores)
self.assertEqual(box_predictor.num_classes, 90)
self.assertTrue(box_predictor._is_training)
def test_construct_default_conv_box_predictor_with_custom_mask_head(self):
box_predictor_text_proto = """
convolutional_box_predictor {
mask_head {
mask_height: 7
mask_width: 7
masks_are_class_agnostic: false
}
conv_hyperparams {
regularizer {
l1_regularizer {
}
}
initializer {
truncated_normal_initializer {
}
}
}
}"""
box_predictor_proto = box_predictor_pb2.BoxPredictor()
text_format.Merge(box_predictor_text_proto, box_predictor_proto)
box_predictor = box_predictor_builder.build(
argscope_fn=hyperparams_builder.build,
box_predictor_config=box_predictor_proto,
is_training=True,
num_classes=90)
self.assertTrue(convolutional_box_predictor.MASK_PREDICTIONS in
box_predictor._other_heads)
mask_prediction_head = (
box_predictor._other_heads[convolutional_box_predictor.MASK_PREDICTIONS]
)
self.assertEqual(mask_prediction_head._mask_height, 7)
self.assertEqual(mask_prediction_head._mask_width, 7)
self.assertFalse(mask_prediction_head._masks_are_class_agnostic)
def test_build_box_predictor_with_mask_branch(self):
box_predictor_proto = box_predictor_pb2.BoxPredictor()
box_predictor_proto.mask_rcnn_box_predictor.fc_hyperparams.op = (
hyperparams_pb2.Hyperparams.FC)
box_predictor_proto.mask_rcnn_box_predictor.conv_hyperparams.op = (
hyperparams_pb2.Hyperparams.CONV)
box_predictor_proto.mask_rcnn_box_predictor.predict_instance_masks = True
box_predictor_proto.mask_rcnn_box_predictor.mask_prediction_conv_depth = 512
mock_argscope_fn = mock.Mock(return_value='arg_scope')
box_predictor = box_predictor_builder.build(
argscope_fn=mock_argscope_fn,
box_predictor_config=box_predictor_proto,
is_training=True,
num_classes=90)
mock_argscope_fn.assert_has_calls(
[mock.call(box_predictor_proto.mask_rcnn_box_predictor.fc_hyperparams,
True),
mock.call(box_predictor_proto.mask_rcnn_box_predictor.conv_hyperparams,
True)], any_order=True)
self.assertFalse(box_predictor._use_dropout)
self.assertAlmostEqual(box_predictor._dropout_keep_prob, 0.5)
self.assertEqual(box_predictor.num_classes, 90)
self.assertTrue(box_predictor._is_training)
self.assertEqual(box_predictor._box_code_size, 4)
self.assertTrue(box_predictor._predict_instance_masks)
self.assertEqual(box_predictor._mask_prediction_conv_depth, 512)
self.assertFalse(box_predictor._predict_keypoints)
def test_box_predictor_builder_calls_fc_argscope_fn(self):
fc_hyperparams_text_proto = """
regularizer {
l1_regularizer {
weight: 0.0003
}
}
initializer {
truncated_normal_initializer {
mean: 0.0
stddev: 0.3
}
}
activation: RELU_6
op: FC
"""
hyperparams_proto = hyperparams_pb2.Hyperparams()
text_format.Merge(fc_hyperparams_text_proto, hyperparams_proto)
box_predictor_proto = box_predictor_pb2.BoxPredictor()
box_predictor_proto.mask_rcnn_box_predictor.fc_hyperparams.CopyFrom(
hyperparams_proto)
mock_argscope_fn = mock.Mock(return_value='arg_scope')
box_predictor = box_predictor_builder.build(
argscope_fn=mock_argscope_fn,
box_predictor_config=box_predictor_proto,
is_training=False,
num_classes=10)
mock_argscope_fn.assert_called_with(hyperparams_proto, False)
self.assertEqual(box_predictor._fc_hyperparams, 'arg_scope')
def test_construct_weight_shared_predictor_with_default_mask_head(self):
box_predictor_text_proto = """
weight_shared_convolutional_box_predictor {
mask_head {
}
conv_hyperparams {
regularizer {
l1_regularizer {
}
}
initializer {
truncated_normal_initializer {
}
}
}
}"""
box_predictor_proto = box_predictor_pb2.BoxPredictor()
text_format.Merge(box_predictor_text_proto, box_predictor_proto)
box_predictor = box_predictor_builder.build(
argscope_fn=hyperparams_builder.build,
box_predictor_config=box_predictor_proto,
is_training=True,
num_classes=90)
self.assertTrue(convolutional_box_predictor.MASK_PREDICTIONS in
box_predictor._other_heads)
weight_shared_convolutional_mask_head = (
box_predictor._other_heads[convolutional_box_predictor.MASK_PREDICTIONS]
)
self.assertIsInstance(weight_shared_convolutional_mask_head,
mask_head.WeightSharedConvolutionalMaskHead)
self.assertEqual(weight_shared_convolutional_mask_head._mask_height, 15)
self.assertEqual(weight_shared_convolutional_mask_head._mask_width, 15)
self.assertTrue(
weight_shared_convolutional_mask_head._masks_are_class_agnostic)
def test_construct_default_conv_box_predictor_with_batch_norm(self):
box_predictor_text_proto = """
weight_shared_convolutional_box_predictor {
conv_hyperparams {
regularizer {
l1_regularizer {
}
}
batch_norm {
train: true
}
initializer {
truncated_normal_initializer {
}
}
}
}"""
box_predictor_proto = box_predictor_pb2.BoxPredictor()
text_format.Merge(box_predictor_text_proto, box_predictor_proto)
box_predictor = box_predictor_builder.build(
argscope_fn=hyperparams_builder.build,
box_predictor_config=box_predictor_proto,
is_training=True,
num_classes=90)
self.assertEqual(box_predictor._depth, 0)
self.assertEqual(box_predictor._num_layers_before_predictor, 0)
self.assertEqual(box_predictor.num_classes, 90)
self.assertTrue(box_predictor._is_training)
self.assertEqual(box_predictor._apply_batch_norm, True)
def test_default_rfcn_box_predictor(self):
conv_hyperparams_text_proto = """
regularizer {
l1_regularizer {
}
}
initializer {
truncated_normal_initializer {
}
}
activation: RELU_6
"""
hyperparams_proto = hyperparams_pb2.Hyperparams()
text_format.Merge(conv_hyperparams_text_proto, hyperparams_proto)
def mock_conv_argscope_builder(conv_hyperparams_arg, is_training):
return (conv_hyperparams_arg, is_training)
box_predictor_proto = box_predictor_pb2.BoxPredictor()
box_predictor_proto.rfcn_box_predictor.conv_hyperparams.CopyFrom(
hyperparams_proto)
box_predictor = box_predictor_builder.build(
argscope_fn=mock_conv_argscope_builder,
box_predictor_config=box_predictor_proto,
is_training=True,
num_classes=90)
self.assertEqual(box_predictor.num_classes, 90)
self.assertTrue(box_predictor._is_training)
self.assertEqual(box_predictor._box_code_size, 4)
self.assertEqual(box_predictor._num_spatial_bins, [3, 3])
self.assertEqual(box_predictor._crop_size, [12, 12])
def _get_second_stage_box_predictor(self, num_classes, is_training):
box_predictor_proto = box_predictor_pb2.BoxPredictor()
text_format.Merge(self._get_second_stage_box_predictor_text_proto(),
box_predictor_proto)
return box_predictor_builder.build(
hyperparams_builder.build,
box_predictor_proto,
num_classes=num_classes,
is_training=is_training)
def _build_ssd_model(ssd_config, is_training, add_summaries):
"""Builds an SSD detection model based on the model config.
Args:
ssd_config: A ssd.proto object containing the config for the desired
SSDMetaArch.
is_training: True if this model is being built for training purposes.
add_summaries: Whether to add tf summaries in the model.
Returns:
SSDMetaArch based on the config.
Raises:
ValueError: If ssd_config.type is not recognized (i.e. not registered in
model_class_map).
"""
num_classes = ssd_config.num_classes
# Feature extractor
feature_extractor = _build_ssd_feature_extractor(ssd_config.feature_extractor,
is_training)
box_coder = box_coder_builder.build(ssd_config.box_coder)
matcher = matcher_builder.build(ssd_config.matcher)
region_similarity_calculator = sim_calc.build(
ssd_config.similarity_calculator)
ssd_box_predictor = box_predictor_builder.build(hyperparams_builder.build,
ssd_config.box_predictor,
is_training, num_classes)
anchor_generator = anchor_generator_builder.build(
ssd_config.anchor_generator)
image_resizer_fn = image_resizer_builder.build(ssd_config.image_resizer)
non_max_suppression_fn, score_conversion_fn = post_processing_builder.build(
ssd_config.post_processing)
(classification_loss, localization_loss, classification_weight,
localization_weight,
hard_example_miner) = losses_builder.build(ssd_config.loss)
normalize_loss_by_num_matches = ssd_config.normalize_loss_by_num_matches
return ssd_meta_arch.SSDMetaArch(
is_training,
anchor_generator,
ssd_box_predictor,
box_coder,
feature_extractor,
matcher,
region_similarity_calculator,
image_resizer_fn,
non_max_suppression_fn,
score_conversion_fn,
classification_loss,
localization_loss,
classification_weight,
localization_weight,
normalize_loss_by_num_matches,
hard_example_miner,
add_summaries=add_summaries)
def test_construct_non_default_conv_box_predictor(self):
box_predictor_text_proto = """
convolutional_box_predictor {
min_depth: 2
max_depth: 16
num_layers_before_predictor: 2
use_dropout: false
dropout_keep_probability: 0.4
kernel_size: 3
box_code_size: 3
apply_sigmoid_to_scores: true
class_prediction_bias_init: 4.0
use_depthwise: true
}
"""
conv_hyperparams_text_proto = """
regularizer {
l1_regularizer {
}
}
initializer {
truncated_normal_initializer {
}
}
"""
hyperparams_proto = hyperparams_pb2.Hyperparams()
text_format.Merge(conv_hyperparams_text_proto, hyperparams_proto)
def mock_conv_argscope_builder(conv_hyperparams_arg, is_training):
return (conv_hyperparams_arg, is_training)
box_predictor_proto = box_predictor_pb2.BoxPredictor()
text_format.Merge(box_predictor_text_proto, box_predictor_proto)
box_predictor_proto.convolutional_box_predictor.conv_hyperparams.CopyFrom(
hyperparams_proto)
box_predictor = box_predictor_builder.build(
argscope_fn=mock_conv_argscope_builder,
box_predictor_config=box_predictor_proto,
is_training=False,
num_classes=10,
add_background_class=False)
class_head = box_predictor._class_prediction_head
self.assertEqual(box_predictor._min_depth, 2)
self.assertEqual(box_predictor._max_depth, 16)
self.assertEqual(box_predictor._num_layers_before_predictor, 2)
self.assertFalse(class_head._use_dropout)
self.assertAlmostEqual(class_head._dropout_keep_prob, 0.4)
self.assertTrue(class_head._apply_sigmoid_to_scores)
self.assertAlmostEqual(class_head._class_prediction_bias_init, 4.0)
self.assertEqual(class_head._num_class_slots, 10)
self.assertEqual(box_predictor.num_classes, 10)
self.assertFalse(box_predictor._is_training)
self.assertTrue(class_head._use_depthwise)
def test_build_default_mask_rcnn_box_predictor(self):
box_predictor_proto = box_predictor_pb2.BoxPredictor()
box_predictor_proto.mask_rcnn_box_predictor.fc_hyperparams.op = (
hyperparams_pb2.Hyperparams.FC)
box_predictor = box_predictor_builder.build(
argscope_fn=mock.Mock(return_value='arg_scope'),
box_predictor_config=box_predictor_proto,
is_training=True,
num_classes=90)
self.assertFalse(box_predictor._use_dropout)
self.assertAlmostEqual(box_predictor._dropout_keep_prob, 0.5)
self.assertEqual(box_predictor.num_classes, 90)
self.assertTrue(box_predictor._is_training)
self.assertEqual(box_predictor._box_code_size, 4)
self.assertFalse(box_predictor._predict_instance_masks)
self.assertFalse(box_predictor._predict_keypoints)
def test_non_default_mask_rcnn_box_predictor(self):
fc_hyperparams_text_proto = """
regularizer {
l1_regularizer {
}
}
initializer {
truncated_normal_initializer {
}
}
activation: RELU_6
op: FC
"""
box_predictor_text_proto = """
mask_rcnn_box_predictor {
use_dropout: true
dropout_keep_probability: 0.8
box_code_size: 3
share_box_across_classes: true
}
"""
hyperparams_proto = hyperparams_pb2.Hyperparams()
text_format.Merge(fc_hyperparams_text_proto, hyperparams_proto)
def mock_fc_argscope_builder(fc_hyperparams_arg, is_training):
return (fc_hyperparams_arg, is_training)
box_predictor_proto = box_predictor_pb2.BoxPredictor()
text_format.Merge(box_predictor_text_proto, box_predictor_proto)
box_predictor_proto.mask_rcnn_box_predictor.fc_hyperparams.CopyFrom(
hyperparams_proto)
box_predictor = box_predictor_builder.build(
argscope_fn=mock_fc_argscope_builder,
box_predictor_config=box_predictor_proto,
is_training=True,
num_classes=90)
box_head = box_predictor._box_prediction_head
class_head = box_predictor._class_prediction_head
self.assertTrue(box_head._use_dropout)
self.assertTrue(class_head._use_dropout)
self.assertAlmostEqual(box_head._dropout_keep_prob, 0.8)
self.assertAlmostEqual(class_head._dropout_keep_prob, 0.8)
self.assertEqual(box_predictor.num_classes, 90)
self.assertTrue(box_predictor._is_training)
self.assertEqual(box_head._box_code_size, 3)
self.assertEqual(box_head._share_box_across_classes, True)
def test_box_predictor_calls_conv_argscope_fn(self):
conv_hyperparams_text_proto = """
regularizer {
l1_regularizer {
weight: 0.0003
}
}
initializer {
truncated_normal_initializer {
mean: 0.0
stddev: 0.3
}
}
activation: RELU_6
"""
hyperparams_proto = hyperparams_pb2.Hyperparams()
text_format.Merge(conv_hyperparams_text_proto, hyperparams_proto)
def mock_conv_argscope_builder(conv_hyperparams_arg, is_training):
return (conv_hyperparams_arg, is_training)
box_predictor_proto = box_predictor_pb2.BoxPredictor()
box_predictor_proto.convolutional_box_predictor.conv_hyperparams.CopyFrom(
hyperparams_proto)
box_predictor = box_predictor_builder.build(
argscope_fn=mock_conv_argscope_builder,
box_predictor_config=box_predictor_proto,
is_training=False,
num_classes=10)
(conv_hyperparams_actual, is_training) = box_predictor._conv_hyperparams
self.assertAlmostEqual((hyperparams_proto.regularizer.
l1_regularizer.weight),
(conv_hyperparams_actual.regularizer.l1_regularizer.
weight))
self.assertAlmostEqual((hyperparams_proto.initializer.
truncated_normal_initializer.stddev),
(conv_hyperparams_actual.initializer.
truncated_normal_initializer.stddev))
self.assertAlmostEqual((hyperparams_proto.initializer.
truncated_normal_initializer.mean),
(conv_hyperparams_actual.initializer.
truncated_normal_initializer.mean))
self.assertEqual(hyperparams_proto.activation,
conv_hyperparams_actual.activation)
self.assertFalse(is_training)
def test_construct_non_default_depthwise_conv_box_predictor(self):
box_predictor_text_proto = """
weight_shared_convolutional_box_predictor {
depth: 2
num_layers_before_predictor: 2
kernel_size: 7
box_code_size: 3
class_prediction_bias_init: 4.0
use_depthwise: true
}
"""
conv_hyperparams_text_proto = """
regularizer {
l1_regularizer {
}
}
initializer {
truncated_normal_initializer {
}
}
"""
hyperparams_proto = hyperparams_pb2.Hyperparams()
text_format.Merge(conv_hyperparams_text_proto, hyperparams_proto)
def mock_conv_argscope_builder(conv_hyperparams_arg, is_training):
return (conv_hyperparams_arg, is_training)
box_predictor_proto = box_predictor_pb2.BoxPredictor()
text_format.Merge(box_predictor_text_proto, box_predictor_proto)
(box_predictor_proto.weight_shared_convolutional_box_predictor.
conv_hyperparams.CopyFrom(hyperparams_proto))
box_predictor = box_predictor_builder.build(
argscope_fn=mock_conv_argscope_builder,
box_predictor_config=box_predictor_proto,
is_training=False,
num_classes=10,
add_background_class=False)
class_head = box_predictor._class_prediction_head
self.assertEqual(box_predictor._depth, 2)
self.assertEqual(box_predictor._num_layers_before_predictor, 2)
self.assertEqual(box_predictor._apply_batch_norm, False)
self.assertEqual(box_predictor._use_depthwise, True)
self.assertAlmostEqual(class_head._class_prediction_bias_init, 4.0)
self.assertEqual(box_predictor.num_classes, 10)
self.assertFalse(box_predictor._is_training)
def test_build_box_predictor_with_convlve_then_upsample_masks(self):
box_predictor_proto = box_predictor_pb2.BoxPredictor()
box_predictor_proto.mask_rcnn_box_predictor.fc_hyperparams.op = (
hyperparams_pb2.Hyperparams.FC)
box_predictor_proto.mask_rcnn_box_predictor.conv_hyperparams.op = (
hyperparams_pb2.Hyperparams.CONV)
box_predictor_proto.mask_rcnn_box_predictor.predict_instance_masks = True
box_predictor_proto.mask_rcnn_box_predictor.mask_prediction_conv_depth = 512
box_predictor_proto.mask_rcnn_box_predictor.mask_height = 24
box_predictor_proto.mask_rcnn_box_predictor.mask_width = 24
box_predictor_proto.mask_rcnn_box_predictor.convolve_then_upsample_masks = (
True)
mock_argscope_fn = mock.Mock(return_value='arg_scope')
box_predictor = box_predictor_builder.build(
argscope_fn=mock_argscope_fn,
box_predictor_config=box_predictor_proto,
is_training=True,
num_classes=90)
mock_argscope_fn.assert_has_calls(
[mock.call(box_predictor_proto.mask_rcnn_box_predictor.fc_hyperparams,
True),
mock.call(box_predictor_proto.mask_rcnn_box_predictor.conv_hyperparams,
True)], any_order=True)
box_head = box_predictor._box_prediction_head
class_head = box_predictor._class_prediction_head
third_stage_heads = box_predictor._third_stage_heads
self.assertFalse(box_head._use_dropout)
self.assertFalse(class_head._use_dropout)
self.assertAlmostEqual(box_head._dropout_keep_prob, 0.5)
self.assertAlmostEqual(class_head._dropout_keep_prob, 0.5)
self.assertEqual(box_predictor.num_classes, 90)
self.assertTrue(box_predictor._is_training)
self.assertEqual(box_head._box_code_size, 4)
self.assertTrue(
mask_rcnn_box_predictor.MASK_PREDICTIONS in third_stage_heads)
self.assertEqual(
third_stage_heads[mask_rcnn_box_predictor.MASK_PREDICTIONS]
._mask_prediction_conv_depth, 512)
self.assertTrue(third_stage_heads[mask_rcnn_box_predictor.MASK_PREDICTIONS]
._convolve_then_upsample)
def test_build_box_predictor_with_mask_branch(self):
box_predictor_proto = box_predictor_pb2.BoxPredictor()
box_predictor_proto.mask_rcnn_box_predictor.fc_hyperparams.op = (
hyperparams_pb2.Hyperparams.FC)
box_predictor_proto.mask_rcnn_box_predictor.conv_hyperparams.op = (
hyperparams_pb2.Hyperparams.CONV)
box_predictor_proto.mask_rcnn_box_predictor.predict_instance_masks = True
box_predictor_proto.mask_rcnn_box_predictor.mask_prediction_conv_depth = 512
box_predictor_proto.mask_rcnn_box_predictor.mask_height = 16
box_predictor_proto.mask_rcnn_box_predictor.mask_width = 16
mock_argscope_fn = mock.Mock(return_value='arg_scope')
box_predictor = box_predictor_builder.build(
argscope_fn=mock_argscope_fn,
box_predictor_config=box_predictor_proto,
is_training=True,
num_classes=90)
mock_argscope_fn.assert_has_calls([
mock.call(
box_predictor_proto.mask_rcnn_box_predictor.fc_hyperparams,
True),
mock.call(
box_predictor_proto.mask_rcnn_box_predictor.conv_hyperparams,
True)
],
any_order=True)
box_head = box_predictor._box_prediction_head
class_head = box_predictor._class_prediction_head
third_stage_heads = box_predictor._third_stage_heads
self.assertFalse(box_head._use_dropout)
self.assertFalse(class_head._use_dropout)
self.assertAlmostEqual(box_head._dropout_keep_prob, 0.5)
self.assertAlmostEqual(class_head._dropout_keep_prob, 0.5)
self.assertEqual(box_predictor.num_classes, 90)
self.assertTrue(box_predictor._is_training)
self.assertEqual(box_head._box_code_size, 4)
self.assertIn(mask_rcnn_box_predictor.MASK_PREDICTIONS,
third_stage_heads)
self.assertEqual(
third_stage_heads[mask_rcnn_box_predictor.MASK_PREDICTIONS].
_mask_prediction_conv_depth, 512)
def test_non_default_mask_rcnn_box_predictor(self):
fc_hyperparams_text_proto = """
regularizer {
l1_regularizer {
}
}
initializer {
truncated_normal_initializer {
}
}
activation: RELU_6
op: FC
"""
box_predictor_text_proto = """
mask_rcnn_box_predictor {
use_dropout: true
dropout_keep_probability: 0.8
box_code_size: 3
}
"""
hyperparams_proto = hyperparams_pb2.Hyperparams()
text_format.Merge(fc_hyperparams_text_proto, hyperparams_proto)
def mock_fc_argscope_builder(fc_hyperparams_arg, is_training):
return (fc_hyperparams_arg, is_training)
box_predictor_proto = box_predictor_pb2.BoxPredictor()
text_format.Merge(box_predictor_text_proto, box_predictor_proto)
box_predictor_proto.mask_rcnn_box_predictor.fc_hyperparams.CopyFrom(
hyperparams_proto)
box_predictor = box_predictor_builder.build(
argscope_fn=mock_fc_argscope_builder,
box_predictor_config=box_predictor_proto,
is_training=True,
num_classes=90)
self.assertTrue(box_predictor._use_dropout)
self.assertAlmostEqual(box_predictor._dropout_keep_prob, 0.8)
self.assertEqual(box_predictor.num_classes, 90)
self.assertTrue(box_predictor._is_training)
self.assertEqual(box_predictor._box_code_size, 3)
def test_construct_non_default_conv_box_predictor(self):
box_predictor_text_proto = """
weight_shared_convolutional_box_predictor {
depth: 2
num_layers_before_predictor: 2
kernel_size: 7
box_code_size: 3
class_prediction_bias_init: 4.0
}
"""
conv_hyperparams_text_proto = """
regularizer {
l1_regularizer {
}
}
initializer {
truncated_normal_initializer {
}
}
"""
hyperparams_proto = hyperparams_pb2.Hyperparams()
text_format.Merge(conv_hyperparams_text_proto, hyperparams_proto)
def mock_conv_argscope_builder(conv_hyperparams_arg, is_training):
return (conv_hyperparams_arg, is_training)
box_predictor_proto = box_predictor_pb2.BoxPredictor()
text_format.Merge(box_predictor_text_proto, box_predictor_proto)
(box_predictor_proto.weight_shared_convolutional_box_predictor.
conv_hyperparams.CopyFrom(hyperparams_proto))
box_predictor = box_predictor_builder.build(
argscope_fn=mock_conv_argscope_builder,
box_predictor_config=box_predictor_proto,
is_training=False,
num_classes=10)
self.assertEqual(box_predictor._depth, 2)
self.assertEqual(box_predictor._num_layers_before_predictor, 2)
self.assertAlmostEqual(box_predictor._class_prediction_bias_init, 4.0)
self.assertEqual(box_predictor.num_classes, 10)
self.assertFalse(box_predictor._is_training)
def test_construct_weight_shared_predictor_with_custom_mask_head(self):
box_predictor_text_proto = """
weight_shared_convolutional_box_predictor {
mask_head {
mask_height: 7
mask_width: 7
masks_are_class_agnostic: false
}
conv_hyperparams {
regularizer {
l1_regularizer {
}
}
initializer {
truncated_normal_initializer {
}
}
}
}"""
box_predictor_proto = box_predictor_pb2.BoxPredictor()
text_format.Merge(box_predictor_text_proto, box_predictor_proto)
box_predictor = box_predictor_builder.build(
argscope_fn=hyperparams_builder.build,
box_predictor_config=box_predictor_proto,
is_training=True,
num_classes=90)
self.assertTrue(convolutional_box_predictor.MASK_PREDICTIONS in
box_predictor._other_heads)
weight_shared_convolutional_mask_head = (box_predictor._other_heads[
convolutional_box_predictor.MASK_PREDICTIONS])
self.assertIsInstance(weight_shared_convolutional_mask_head,
mask_head.WeightSharedConvolutionalMaskHead)
self.assertEqual(weight_shared_convolutional_mask_head._mask_height, 7)
self.assertEqual(weight_shared_convolutional_mask_head._mask_width, 7)
self.assertFalse(
weight_shared_convolutional_mask_head._masks_are_class_agnostic)
def _build_faster_rcnn_model(frcnn_config, is_training, add_summaries):
"""Builds a Faster R-CNN or R-FCN detection model based on the model config.
Builds R-FCN model if the second_stage_box_predictor in the config is of type
`rfcn_box_predictor` else builds a Faster R-CNN model.
Args:
frcnn_config: A faster_rcnn.proto object containing the config for the
desired FasterRCNNMetaArch or RFCNMetaArch.
is_training: True if this model is being built for training purposes.
add_summaries: Whether to add tf summaries in the model.
Returns:
FasterRCNNMetaArch based on the config.
Raises:
ValueError: If frcnn_config.type is not recognized (i.e. not registered in
model_class_map).
"""
num_classes = frcnn_config.num_classes
image_resizer_fn = image_resizer_builder.build(frcnn_config.image_resizer)
is_keras = (frcnn_config.feature_extractor.type
in FASTER_RCNN_KERAS_FEATURE_EXTRACTOR_CLASS_MAP)
if is_keras:
feature_extractor = _build_faster_rcnn_keras_feature_extractor(
frcnn_config.feature_extractor,
is_training,
inplace_batchnorm_update=frcnn_config.inplace_batchnorm_update)
else:
feature_extractor = _build_faster_rcnn_feature_extractor(
frcnn_config.feature_extractor,
is_training,
inplace_batchnorm_update=frcnn_config.inplace_batchnorm_update)
number_of_stages = frcnn_config.number_of_stages
first_stage_anchor_generator = anchor_generator_builder.build(
frcnn_config.first_stage_anchor_generator)
first_stage_target_assigner = target_assigner.create_target_assigner(
'FasterRCNN',
'proposal',
use_matmul_gather=frcnn_config.use_matmul_gather_in_matcher)
first_stage_atrous_rate = frcnn_config.first_stage_atrous_rate
if is_keras:
first_stage_box_predictor_arg_scope_fn = (
hyperparams_builder.KerasLayerHyperparams(
frcnn_config.first_stage_box_predictor_conv_hyperparams))
else:
first_stage_box_predictor_arg_scope_fn = hyperparams_builder.build(
frcnn_config.first_stage_box_predictor_conv_hyperparams,
is_training)
first_stage_box_predictor_kernel_size = (
frcnn_config.first_stage_box_predictor_kernel_size)
first_stage_box_predictor_depth = frcnn_config.first_stage_box_predictor_depth
first_stage_minibatch_size = frcnn_config.first_stage_minibatch_size
use_static_shapes = frcnn_config.use_static_shapes and (
frcnn_config.use_static_shapes_for_eval or is_training)
first_stage_sampler = sampler.BalancedPositiveNegativeSampler(
positive_fraction=frcnn_config.first_stage_positive_balance_fraction,
is_static=(frcnn_config.use_static_balanced_label_sampler
and use_static_shapes))
first_stage_max_proposals = frcnn_config.first_stage_max_proposals
if (frcnn_config.first_stage_nms_iou_threshold < 0
or frcnn_config.first_stage_nms_iou_threshold > 1.0):
raise ValueError('iou_threshold not in [0, 1.0].')
if (is_training and
frcnn_config.second_stage_batch_size > first_stage_max_proposals):
raise ValueError('second_stage_batch_size should be no greater than '
'first_stage_max_proposals.')
first_stage_non_max_suppression_fn = functools.partial(
post_processing.batch_multiclass_non_max_suppression,
score_thresh=frcnn_config.first_stage_nms_score_threshold,
iou_thresh=frcnn_config.first_stage_nms_iou_threshold,
max_size_per_class=frcnn_config.first_stage_max_proposals,
max_total_size=frcnn_config.first_stage_max_proposals,
use_static_shapes=use_static_shapes,
use_partitioned_nms=frcnn_config.use_partitioned_nms_in_first_stage,
use_combined_nms=frcnn_config.use_combined_nms_in_first_stage)
first_stage_loc_loss_weight = (
frcnn_config.first_stage_localization_loss_weight)
first_stage_obj_loss_weight = frcnn_config.first_stage_objectness_loss_weight
initial_crop_size = frcnn_config.initial_crop_size
maxpool_kernel_size = frcnn_config.maxpool_kernel_size
maxpool_stride = frcnn_config.maxpool_stride
second_stage_target_assigner = target_assigner.create_target_assigner(
'FasterRCNN',
'detection',
use_matmul_gather=frcnn_config.use_matmul_gather_in_matcher)
if is_keras:
second_stage_box_predictor = box_predictor_builder.build_keras(
hyperparams_builder.KerasLayerHyperparams,
freeze_batchnorm=False,
inplace_batchnorm_update=False,
num_predictions_per_location_list=[1],
box_predictor_config=frcnn_config.second_stage_box_predictor,
is_training=is_training,
num_classes=num_classes)
else:
second_stage_box_predictor = box_predictor_builder.build(
hyperparams_builder.build,
frcnn_config.second_stage_box_predictor,
is_training=is_training,
num_classes=num_classes)
second_stage_batch_size = frcnn_config.second_stage_batch_size
second_stage_sampler = sampler.BalancedPositiveNegativeSampler(
positive_fraction=frcnn_config.second_stage_balance_fraction,
is_static=(frcnn_config.use_static_balanced_label_sampler
and use_static_shapes))
(second_stage_non_max_suppression_fn,
second_stage_score_conversion_fn) = post_processing_builder.build(
frcnn_config.second_stage_post_processing)
second_stage_localization_loss_weight = (
frcnn_config.second_stage_localization_loss_weight)
second_stage_classification_loss = (
losses_builder.build_faster_rcnn_classification_loss(
frcnn_config.second_stage_classification_loss))
second_stage_classification_loss_weight = (
frcnn_config.second_stage_classification_loss_weight)
second_stage_mask_prediction_loss_weight = (
frcnn_config.second_stage_mask_prediction_loss_weight)
hard_example_miner = None
if frcnn_config.HasField('hard_example_miner'):
hard_example_miner = losses_builder.build_hard_example_miner(
frcnn_config.hard_example_miner,
second_stage_classification_loss_weight,
second_stage_localization_loss_weight)
crop_and_resize_fn = (ops.matmul_crop_and_resize
if frcnn_config.use_matmul_crop_and_resize else
ops.native_crop_and_resize)
clip_anchors_to_image = (frcnn_config.clip_anchors_to_image)
common_kwargs = {
'is_training':
is_training,
'num_classes':
num_classes,
'image_resizer_fn':
image_resizer_fn,
'feature_extractor':
feature_extractor,
'number_of_stages':
number_of_stages,
'first_stage_anchor_generator':
first_stage_anchor_generator,
'first_stage_target_assigner':
first_stage_target_assigner,
'first_stage_atrous_rate':
first_stage_atrous_rate,
'first_stage_box_predictor_arg_scope_fn':
first_stage_box_predictor_arg_scope_fn,
'first_stage_box_predictor_kernel_size':
first_stage_box_predictor_kernel_size,
'first_stage_box_predictor_depth':
first_stage_box_predictor_depth,
'first_stage_minibatch_size':
first_stage_minibatch_size,
'first_stage_sampler':
first_stage_sampler,
'first_stage_non_max_suppression_fn':
first_stage_non_max_suppression_fn,
'first_stage_max_proposals':
first_stage_max_proposals,
'first_stage_localization_loss_weight':
first_stage_loc_loss_weight,
'first_stage_objectness_loss_weight':
first_stage_obj_loss_weight,
'second_stage_target_assigner':
second_stage_target_assigner,
'second_stage_batch_size':
second_stage_batch_size,
'second_stage_sampler':
second_stage_sampler,
'second_stage_non_max_suppression_fn':
second_stage_non_max_suppression_fn,
'second_stage_score_conversion_fn':
second_stage_score_conversion_fn,
'second_stage_localization_loss_weight':
second_stage_localization_loss_weight,
'second_stage_classification_loss':
second_stage_classification_loss,
'second_stage_classification_loss_weight':
second_stage_classification_loss_weight,
'hard_example_miner':
hard_example_miner,
'add_summaries':
add_summaries,
'crop_and_resize_fn':
crop_and_resize_fn,
'clip_anchors_to_image':
clip_anchors_to_image,
'use_static_shapes':
use_static_shapes,
'resize_masks':
frcnn_config.resize_masks,
'return_raw_detections_during_predict':
(frcnn_config.return_raw_detections_during_predict)
}
if (isinstance(second_stage_box_predictor,
rfcn_box_predictor.RfcnBoxPredictor)
or isinstance(second_stage_box_predictor,
rfcn_keras_box_predictor.RfcnKerasBoxPredictor)):
return rfcn_meta_arch.RFCNMetaArch(
second_stage_rfcn_box_predictor=second_stage_box_predictor,
**common_kwargs)
else:
return faster_rcnn_meta_arch.FasterRCNNMetaArch(
initial_crop_size=initial_crop_size,
maxpool_kernel_size=maxpool_kernel_size,
maxpool_stride=maxpool_stride,
second_stage_mask_rcnn_box_predictor=second_stage_box_predictor,
second_stage_mask_prediction_loss_weight=(
second_stage_mask_prediction_loss_weight),
**common_kwargs)
def _build_faster_rcnn_model(frcnn_config, is_training, add_summaries):
"""Builds a Faster R-CNN or R-FCN detection model based on the model config.
Builds R-FCN model if the second_stage_box_predictor in the config is of type
`rfcn_box_predictor` else builds a Faster R-CNN model.
Args:
frcnn_config: A faster_rcnn.proto object containing the config for the
desired FasterRCNNMetaArch or RFCNMetaArch.
is_training: True if this model is being built for training purposes.
add_summaries: Whether to add tf summaries in the model.
Returns:
FasterRCNNMetaArch based on the config.
Raises:
ValueError: If frcnn_config.type is not recognized (i.e. not registered in
model_class_map).
"""
num_classes = frcnn_config.num_classes
image_resizer_fn = image_resizer_builder.build(frcnn_config.image_resizer)
feature_extractor = _build_faster_rcnn_feature_extractor(
frcnn_config.feature_extractor, is_training,
frcnn_config.inplace_batchnorm_update)
number_of_stages = frcnn_config.number_of_stages
first_stage_anchor_generator = anchor_generator_builder.build(
frcnn_config.first_stage_anchor_generator)
first_stage_atrous_rate = frcnn_config.first_stage_atrous_rate
first_stage_box_predictor_arg_scope_fn = hyperparams_builder.build(
frcnn_config.first_stage_box_predictor_conv_hyperparams, is_training)
first_stage_box_predictor_kernel_size = (
frcnn_config.first_stage_box_predictor_kernel_size)
first_stage_box_predictor_depth = frcnn_config.first_stage_box_predictor_depth
first_stage_minibatch_size = frcnn_config.first_stage_minibatch_size
first_stage_positive_balance_fraction = (
frcnn_config.first_stage_positive_balance_fraction)
first_stage_nms_score_threshold = frcnn_config.first_stage_nms_score_threshold
first_stage_nms_iou_threshold = frcnn_config.first_stage_nms_iou_threshold
first_stage_max_proposals = frcnn_config.first_stage_max_proposals
first_stage_loc_loss_weight = (
frcnn_config.first_stage_localization_loss_weight)
first_stage_obj_loss_weight = frcnn_config.first_stage_objectness_loss_weight
initial_crop_size = frcnn_config.initial_crop_size
maxpool_kernel_size = frcnn_config.maxpool_kernel_size
maxpool_stride = frcnn_config.maxpool_stride
second_stage_box_predictor = box_predictor_builder.build(
hyperparams_builder.build,
frcnn_config.second_stage_box_predictor,
is_training=is_training,
num_classes=num_classes)
second_stage_batch_size = frcnn_config.second_stage_batch_size
second_stage_balance_fraction = frcnn_config.second_stage_balance_fraction
(second_stage_non_max_suppression_fn,
second_stage_score_conversion_fn) = post_processing_builder.build(
frcnn_config.second_stage_post_processing)
second_stage_localization_loss_weight = (
frcnn_config.second_stage_localization_loss_weight)
second_stage_classification_loss = (
losses_builder.build_faster_rcnn_classification_loss(
frcnn_config.second_stage_classification_loss))
second_stage_classification_loss_weight = (
frcnn_config.second_stage_classification_loss_weight)
second_stage_mask_prediction_loss_weight = (
frcnn_config.second_stage_mask_prediction_loss_weight)
hard_example_miner = None
if frcnn_config.HasField('hard_example_miner'):
hard_example_miner = losses_builder.build_hard_example_miner(
frcnn_config.hard_example_miner,
second_stage_classification_loss_weight,
second_stage_localization_loss_weight)
common_kwargs = {
'is_training': is_training,
'num_classes': num_classes,
'image_resizer_fn': image_resizer_fn,
'feature_extractor': feature_extractor,
'number_of_stages': number_of_stages,
'first_stage_anchor_generator': first_stage_anchor_generator,
'first_stage_atrous_rate': first_stage_atrous_rate,
'first_stage_box_predictor_arg_scope_fn':
first_stage_box_predictor_arg_scope_fn,
'first_stage_box_predictor_kernel_size':
first_stage_box_predictor_kernel_size,
'first_stage_box_predictor_depth': first_stage_box_predictor_depth,
'first_stage_minibatch_size': first_stage_minibatch_size,
'first_stage_positive_balance_fraction':
first_stage_positive_balance_fraction,
'first_stage_nms_score_threshold': first_stage_nms_score_threshold,
'first_stage_nms_iou_threshold': first_stage_nms_iou_threshold,
'first_stage_max_proposals': first_stage_max_proposals,
'first_stage_localization_loss_weight': first_stage_loc_loss_weight,
'first_stage_objectness_loss_weight': first_stage_obj_loss_weight,
'second_stage_batch_size': second_stage_batch_size,
'second_stage_balance_fraction': second_stage_balance_fraction,
'second_stage_non_max_suppression_fn':
second_stage_non_max_suppression_fn,
'second_stage_score_conversion_fn': second_stage_score_conversion_fn,
'second_stage_localization_loss_weight':
second_stage_localization_loss_weight,
'second_stage_classification_loss': second_stage_classification_loss,
'second_stage_classification_loss_weight':
second_stage_classification_loss_weight,
'hard_example_miner': hard_example_miner,
'add_summaries': add_summaries
}
if isinstance(second_stage_box_predictor, box_predictor.RfcnBoxPredictor):
return rfcn_meta_arch.RFCNMetaArch(
second_stage_rfcn_box_predictor=second_stage_box_predictor,
**common_kwargs)
else:
return faster_rcnn_meta_arch.FasterRCNNMetaArch(
initial_crop_size=initial_crop_size,
maxpool_kernel_size=maxpool_kernel_size,
maxpool_stride=maxpool_stride,
second_stage_mask_rcnn_box_predictor=second_stage_box_predictor,
second_stage_mask_prediction_loss_weight=(
second_stage_mask_prediction_loss_weight),
**common_kwargs)
def _build_ssd_model(ssd_config, is_training, add_summaries):
"""Builds an SSD detection model based on the model config.
Args:
ssd_config: A ssd.proto object containing the config for the desired
SSDMetaArch.
is_training: True if this model is being built for training purposes.
add_summaries: Whether to add tf summaries in the model.
Returns:
SSDMetaArch based on the config.
Raises:
ValueError: If ssd_config.type is not recognized (i.e. not registered in
model_class_map).
"""
num_classes = ssd_config.num_classes
# Feature extractor
feature_extractor = _build_ssd_feature_extractor(
feature_extractor_config=ssd_config.feature_extractor,
freeze_batchnorm=ssd_config.freeze_batchnorm,
is_training=is_training)
box_coder = box_coder_builder.build(ssd_config.box_coder)
matcher = matcher_builder.build(ssd_config.matcher)
region_similarity_calculator = sim_calc.build(
ssd_config.similarity_calculator)
encode_background_as_zeros = ssd_config.encode_background_as_zeros
negative_class_weight = ssd_config.negative_class_weight
anchor_generator = anchor_generator_builder.build(
ssd_config.anchor_generator)
if feature_extractor.is_keras_model:
ssd_box_predictor = box_predictor_builder.build_keras(
hyperparams_fn=hyperparams_builder.KerasLayerHyperparams,
freeze_batchnorm=ssd_config.freeze_batchnorm,
inplace_batchnorm_update=False,
num_predictions_per_location_list=anchor_generator.
num_anchors_per_location(),
box_predictor_config=ssd_config.box_predictor,
is_training=is_training,
num_classes=num_classes,
add_background_class=ssd_config.add_background_class)
else:
ssd_box_predictor = box_predictor_builder.build(
hyperparams_builder.build, ssd_config.box_predictor, is_training,
num_classes, ssd_config.add_background_class)
image_resizer_fn = image_resizer_builder.build(ssd_config.image_resizer)
non_max_suppression_fn, score_conversion_fn = post_processing_builder.build(
ssd_config.post_processing)
(classification_loss, localization_loss, classification_weight,
localization_weight, hard_example_miner, random_example_sampler,
expected_loss_weights_fn) = losses_builder.build(ssd_config.loss)
normalize_loss_by_num_matches = ssd_config.normalize_loss_by_num_matches
normalize_loc_loss_by_codesize = ssd_config.normalize_loc_loss_by_codesize
equalization_loss_config = ops.EqualizationLossConfig(
weight=ssd_config.loss.equalization_loss.weight,
exclude_prefixes=ssd_config.loss.equalization_loss.exclude_prefixes)
target_assigner_instance = target_assigner.TargetAssigner(
region_similarity_calculator,
matcher,
box_coder,
negative_class_weight=negative_class_weight)
ssd_meta_arch_fn = ssd_meta_arch.SSDMetaArch
kwargs = {}
return ssd_meta_arch_fn(
is_training=is_training,
anchor_generator=anchor_generator,
box_predictor=ssd_box_predictor,
box_coder=box_coder,
feature_extractor=feature_extractor,
encode_background_as_zeros=encode_background_as_zeros,
image_resizer_fn=image_resizer_fn,
non_max_suppression_fn=non_max_suppression_fn,
score_conversion_fn=score_conversion_fn,
classification_loss=classification_loss,
localization_loss=localization_loss,
classification_loss_weight=classification_weight,
localization_loss_weight=localization_weight,
normalize_loss_by_num_matches=normalize_loss_by_num_matches,
hard_example_miner=hard_example_miner,
target_assigner_instance=target_assigner_instance,
add_summaries=add_summaries,
normalize_loc_loss_by_codesize=normalize_loc_loss_by_codesize,
freeze_batchnorm=ssd_config.freeze_batchnorm,
inplace_batchnorm_update=ssd_config.inplace_batchnorm_update,
add_background_class=ssd_config.add_background_class,
explicit_background_class=ssd_config.explicit_background_class,
random_example_sampler=random_example_sampler,
expected_loss_weights_fn=expected_loss_weights_fn,
use_confidences_as_targets=ssd_config.use_confidences_as_targets,
implicit_example_weight=ssd_config.implicit_example_weight,
equalization_loss_config=equalization_loss_config,
return_raw_detections_during_predict=(
ssd_config.return_raw_detections_during_predict),
**kwargs)
def _build_ssd_model(ssd_config, is_training):
"""Builds an SSD detection model based on the model config.
Args:
ssd_config: A ssd.proto object containing the config for the desired
SSDMetaArch.
is_training: True if this model is being built for training purposes.
Returns:
SSDMetaArch based on the config.
Raises:
ValueError: If ssd_config.type is not recognized (i.e. not registered in
model_class_map).
"""
num_classes = ssd_config.num_classes
# Feature extractor
feature_extractor = _build_ssd_feature_extractor(
ssd_config.feature_extractor, is_training)
box_coder = box_coder_builder.build(ssd_config.box_coder)
matcher = matcher_builder.build(ssd_config.matcher)
region_similarity_calculator = sim_calc.build(
ssd_config.similarity_calculator)
ssd_box_predictor = box_predictor_builder.build(hyperparams_builder.build,
ssd_config.box_predictor,
is_training, num_classes)
anchor_generator = anchor_generator_builder.build(
ssd_config.anchor_generator)
image_resizer_fn = image_resizer_builder.build(ssd_config.image_resizer)
non_max_suppression_fn, score_conversion_fn = post_processing_builder.build(
ssd_config.post_processing)
(classification_loss, localization_loss, classification_weight,
localization_weight,
hard_example_miner) = losses_builder.build(ssd_config.loss)
normalize_loss_by_num_matches = ssd_config.normalize_loss_by_num_matches
common_kwargs = {
'is_training': is_training,
'anchor_generator': anchor_generator,
'box_predictor': ssd_box_predictor,
'box_coder': box_coder,
'feature_extractor': feature_extractor,
'matcher': matcher,
'region_similarity_calculator': region_similarity_calculator,
'image_resizer_fn': image_resizer_fn,
'non_max_suppression_fn': non_max_suppression_fn,
'score_conversion_fn': score_conversion_fn,
'classification_loss': classification_loss,
'localization_loss': localization_loss,
'classification_loss_weight': classification_weight,
'localization_loss_weight': localization_weight,
'normalize_loss_by_num_matches': normalize_loss_by_num_matches,
'hard_example_miner': hard_example_miner
}
if isinstance(anchor_generator,
yolo_grid_anchor_generator.YoloGridAnchorGenerator):
return yolo_meta_arch.YOLOMetaArch(**common_kwargs)
else:
return ssd_meta_arch.SSDMetaArch(**common_kwargs)
def _build_ssd_model(ssd_config,
is_training,
add_summaries,
add_background_class=True):
"""Builds an SSD detection model based on the model config.
Args:
ssd_config: A ssd.proto object containing the config for the desired
SSDMetaArch.
is_training: True if this model is being built for training purposes.
add_summaries: Whether to add tf summaries in the model.
add_background_class: Whether to add an implicit background class to one-hot
encodings of groundtruth labels. Set to false if using groundtruth labels
with an explicit background class or using multiclass scores instead of
truth in the case of distillation.
Returns:
SSDMetaArch based on the config.
Raises:
ValueError: If ssd_config.type is not recognized (i.e. not registered in
model_class_map).
"""
num_classes = ssd_config.num_classes
# Feature extractor
feature_extractor = _build_ssd_feature_extractor(
feature_extractor_config=ssd_config.feature_extractor,
is_training=is_training)
box_coder = box_coder_builder.build(ssd_config.box_coder)
matcher = matcher_builder.build(ssd_config.matcher)
region_similarity_calculator = sim_calc.build(
ssd_config.similarity_calculator)
encode_background_as_zeros = ssd_config.encode_background_as_zeros
negative_class_weight = ssd_config.negative_class_weight
ssd_box_predictor = box_predictor_builder.build(hyperparams_builder.build,
ssd_config.box_predictor,
is_training, num_classes)
anchor_generator = anchor_generator_builder.build(
ssd_config.anchor_generator)
image_resizer_fn = image_resizer_builder.build(ssd_config.image_resizer)
non_max_suppression_fn, score_conversion_fn = post_processing_builder.build(
ssd_config.post_processing)
(classification_loss, localization_loss, classification_weight,
localization_weight, hard_example_miner,
random_example_sampler) = losses_builder.build(ssd_config.loss)
normalize_loss_by_num_matches = ssd_config.normalize_loss_by_num_matches
normalize_loc_loss_by_codesize = ssd_config.normalize_loc_loss_by_codesize
return ssd_meta_arch.SSDMetaArch(
is_training,
anchor_generator,
ssd_box_predictor,
box_coder,
feature_extractor,
matcher,
region_similarity_calculator,
encode_background_as_zeros,
negative_class_weight,
image_resizer_fn,
non_max_suppression_fn,
score_conversion_fn,
classification_loss,
localization_loss,
classification_weight,
localization_weight,
normalize_loss_by_num_matches,
hard_example_miner,
add_summaries=add_summaries,
normalize_loc_loss_by_codesize=normalize_loc_loss_by_codesize,
freeze_batchnorm=ssd_config.freeze_batchnorm,
inplace_batchnorm_update=ssd_config.inplace_batchnorm_update,
add_background_class=add_background_class,
random_example_sampler=random_example_sampler)
def _build_faster_rcnn_model(frcnn_config, is_training, add_summaries):
"""Builds a Faster R-CNN or R-FCN detection model based on the model config.
Builds R-FCN model if the second_stage_box_predictor in the config is of type
`rfcn_box_predictor` else builds a Faster R-CNN model.
Args:
frcnn_config: A faster_rcnn.proto object containing the config for the
desired FasterRCNNMetaArch or RFCNMetaArch.
is_training: True if this model is being built for training purposes.
add_summaries: Whether to add tf summaries in the model.
Returns:
FasterRCNNMetaArch based on the config.
Raises:
ValueError: If frcnn_config.type is not recognized (i.e. not registered in
model_class_map).
"""
num_classes = frcnn_config.num_classes
image_resizer_fn = image_resizer_builder.build(frcnn_config.image_resizer)
feature_extractor = _build_faster_rcnn_feature_extractor(
frcnn_config.feature_extractor, is_training,
frcnn_config.inplace_batchnorm_update)
number_of_stages = frcnn_config.number_of_stages
first_stage_anchor_generator = anchor_generator_builder.build(
frcnn_config.first_stage_anchor_generator)
first_stage_target_assigner = target_assigner.create_target_assigner(
'FasterRCNN',
'proposal',
use_matmul_gather=frcnn_config.use_matmul_gather_in_matcher)
first_stage_atrous_rate = frcnn_config.first_stage_atrous_rate
first_stage_box_predictor_arg_scope_fn = hyperparams_builder.build(
frcnn_config.first_stage_box_predictor_conv_hyperparams, is_training)
first_stage_box_predictor_kernel_size = (
frcnn_config.first_stage_box_predictor_kernel_size)
first_stage_box_predictor_depth = frcnn_config.first_stage_box_predictor_depth
first_stage_minibatch_size = frcnn_config.first_stage_minibatch_size
use_static_shapes = frcnn_config.use_static_shapes
first_stage_sampler = sampler.BalancedPositiveNegativeSampler(
positive_fraction=frcnn_config.first_stage_positive_balance_fraction,
is_static=(frcnn_config.use_static_balanced_label_sampler and
use_static_shapes))
first_stage_max_proposals = frcnn_config.first_stage_max_proposals
if (frcnn_config.first_stage_nms_iou_threshold < 0 or
frcnn_config.first_stage_nms_iou_threshold > 1.0):
raise ValueError('iou_threshold not in [0, 1.0].')
if (is_training and frcnn_config.second_stage_batch_size >
first_stage_max_proposals):
raise ValueError('second_stage_batch_size should be no greater than '
'first_stage_max_proposals.')
first_stage_non_max_suppression_fn = functools.partial(
post_processing.batch_multiclass_non_max_suppression,
score_thresh=frcnn_config.first_stage_nms_score_threshold,
iou_thresh=frcnn_config.first_stage_nms_iou_threshold,
max_size_per_class=frcnn_config.first_stage_max_proposals,
max_total_size=frcnn_config.first_stage_max_proposals,
use_static_shapes=use_static_shapes)
first_stage_loc_loss_weight = (
frcnn_config.first_stage_localization_loss_weight)
first_stage_obj_loss_weight = frcnn_config.first_stage_objectness_loss_weight
initial_crop_size = frcnn_config.initial_crop_size
maxpool_kernel_size = frcnn_config.maxpool_kernel_size
maxpool_stride = frcnn_config.maxpool_stride
second_stage_target_assigner = target_assigner.create_target_assigner(
'FasterRCNN',
'detection',
use_matmul_gather=frcnn_config.use_matmul_gather_in_matcher)
second_stage_box_predictor = box_predictor_builder.build(
hyperparams_builder.build,
frcnn_config.second_stage_box_predictor,
is_training=is_training,
num_classes=num_classes)
second_stage_batch_size = frcnn_config.second_stage_batch_size
second_stage_sampler = sampler.BalancedPositiveNegativeSampler(
positive_fraction=frcnn_config.second_stage_balance_fraction,
is_static=(frcnn_config.use_static_balanced_label_sampler and
use_static_shapes))
(second_stage_non_max_suppression_fn, second_stage_score_conversion_fn
) = post_processing_builder.build(frcnn_config.second_stage_post_processing)
second_stage_localization_loss_weight = (
frcnn_config.second_stage_localization_loss_weight)
second_stage_classification_loss = (
losses_builder.build_faster_rcnn_classification_loss(
frcnn_config.second_stage_classification_loss))
second_stage_classification_loss_weight = (
frcnn_config.second_stage_classification_loss_weight)
second_stage_mask_prediction_loss_weight = (
frcnn_config.second_stage_mask_prediction_loss_weight)
hard_example_miner = None
if frcnn_config.HasField('hard_example_miner'):
hard_example_miner = losses_builder.build_hard_example_miner(
frcnn_config.hard_example_miner,
second_stage_classification_loss_weight,
second_stage_localization_loss_weight)
crop_and_resize_fn = (
ops.matmul_crop_and_resize if frcnn_config.use_matmul_crop_and_resize
else ops.native_crop_and_resize)
clip_anchors_to_image = (
frcnn_config.clip_anchors_to_image)
common_kwargs = {
'is_training': is_training,
'num_classes': num_classes,
'image_resizer_fn': image_resizer_fn,
'feature_extractor': feature_extractor,
'number_of_stages': number_of_stages,
'first_stage_anchor_generator': first_stage_anchor_generator,
'first_stage_target_assigner': first_stage_target_assigner,
'first_stage_atrous_rate': first_stage_atrous_rate,
'first_stage_box_predictor_arg_scope_fn':
first_stage_box_predictor_arg_scope_fn,
'first_stage_box_predictor_kernel_size':
first_stage_box_predictor_kernel_size,
'first_stage_box_predictor_depth': first_stage_box_predictor_depth,
'first_stage_minibatch_size': first_stage_minibatch_size,
'first_stage_sampler': first_stage_sampler,
'first_stage_non_max_suppression_fn': first_stage_non_max_suppression_fn,
'first_stage_max_proposals': first_stage_max_proposals,
'first_stage_localization_loss_weight': first_stage_loc_loss_weight,
'first_stage_objectness_loss_weight': first_stage_obj_loss_weight,
'second_stage_target_assigner': second_stage_target_assigner,
'second_stage_batch_size': second_stage_batch_size,
'second_stage_sampler': second_stage_sampler,
'second_stage_non_max_suppression_fn':
second_stage_non_max_suppression_fn,
'second_stage_score_conversion_fn': second_stage_score_conversion_fn,
'second_stage_localization_loss_weight':
second_stage_localization_loss_weight,
'second_stage_classification_loss':
second_stage_classification_loss,
'second_stage_classification_loss_weight':
second_stage_classification_loss_weight,
'hard_example_miner': hard_example_miner,
'add_summaries': add_summaries,
'crop_and_resize_fn': crop_and_resize_fn,
'clip_anchors_to_image': clip_anchors_to_image,
'use_static_shapes': use_static_shapes,
'resize_masks': frcnn_config.resize_masks
}
if isinstance(second_stage_box_predictor,
rfcn_box_predictor.RfcnBoxPredictor):
return rfcn_meta_arch.RFCNMetaArch(
second_stage_rfcn_box_predictor=second_stage_box_predictor,
**common_kwargs)
else:
return faster_rcnn_meta_arch.FasterRCNNMetaArch(
initial_crop_size=initial_crop_size,
maxpool_kernel_size=maxpool_kernel_size,
maxpool_stride=maxpool_stride,
second_stage_mask_rcnn_box_predictor=second_stage_box_predictor,
second_stage_mask_prediction_loss_weight=(
second_stage_mask_prediction_loss_weight),
**common_kwargs)
def _build_faster_rcnn_model(frcnn_config, is_training, add_summaries, **kwargs):
"""Builds a Faster R-CNN or R-FCN detection model based on the model config.
Builds R-FCN model if the second_stage_box_predictor in the config is of type
`rfcn_box_predictor` else builds a Faster R-CNN model.
Args:
frcnn_config: A faster_rcnn.proto object containing the config for the
desired FasterRCNNMetaArch or RFCNMetaArch.
is_training: True if this model is being built for training purposes.
add_summaries: Whether to add tf summaries in the model.
kwargs: key-value
'rpn_type' is the type of rpn which is 'cascade_rpn','orign_rpn'
and 'without_rpn' which need some boxes replacing the proposal
generated by rpn
'filter_fn_arg' is the args of filter fn which need the boxes to filter
the proposals.
'replace_rpn_arg' is a dictionary.
only if the rpn_type=='without_rpn' and not None, it's useful in order to
replace the proposals generated by rpn with the gt which maybe adjusted.
'type': a string which is 'gt' or 'others'.
'scale': a float which is used to scale the boxes(maybe gt).
Returns:
FasterRCNNMetaArch based on the config.
Raises:
ValueError: If frcnn_config.type is not recognized (i.e. not registered in
model_class_map).
"""
num_classes = frcnn_config.num_classes
image_resizer_fn = image_resizer_builder.build(frcnn_config.image_resizer)
feature_extractor = _build_faster_rcnn_feature_extractor(
frcnn_config.feature_extractor, is_training,
inplace_batchnorm_update=frcnn_config.inplace_batchnorm_update)
number_of_stages = frcnn_config.number_of_stages
first_stage_anchor_generator = anchor_generator_builder.build(
frcnn_config.first_stage_anchor_generator)
first_stage_target_assigner = target_assigner.create_target_assigner(
'FasterRCNN',
'proposal',
use_matmul_gather=frcnn_config.use_matmul_gather_in_matcher)
first_stage_atrous_rate = frcnn_config.first_stage_atrous_rate
first_stage_box_predictor_arg_scope_fn = hyperparams_builder.build(
frcnn_config.first_stage_box_predictor_conv_hyperparams, is_training)
first_stage_box_predictor_kernel_size = (
frcnn_config.first_stage_box_predictor_kernel_size)
first_stage_box_predictor_depth = frcnn_config.first_stage_box_predictor_depth
first_stage_minibatch_size = frcnn_config.first_stage_minibatch_size
use_static_shapes = frcnn_config.use_static_shapes and (
frcnn_config.use_static_shapes_for_eval or is_training)
first_stage_sampler = sampler.BalancedPositiveNegativeSampler(
positive_fraction=frcnn_config.first_stage_positive_balance_fraction,
is_static=(frcnn_config.use_static_balanced_label_sampler and
use_static_shapes))
first_stage_max_proposals = frcnn_config.first_stage_max_proposals
if (frcnn_config.first_stage_nms_iou_threshold < 0 or
frcnn_config.first_stage_nms_iou_threshold > 1.0):
raise ValueError('iou_threshold not in [0, 1.0].')
if (is_training and frcnn_config.second_stage_batch_size >
first_stage_max_proposals):
raise ValueError('second_stage_batch_size should be no greater than '
'first_stage_max_proposals.')
first_stage_non_max_suppression_fn = functools.partial(
post_processing.batch_multiclass_non_max_suppression,
score_thresh=frcnn_config.first_stage_nms_score_threshold,
iou_thresh=frcnn_config.first_stage_nms_iou_threshold,
max_size_per_class=frcnn_config.first_stage_max_proposals,
max_total_size=frcnn_config.first_stage_max_proposals,
use_static_shapes=use_static_shapes)
first_stage_loc_loss_weight = (
frcnn_config.first_stage_localization_loss_weight)
first_stage_obj_loss_weight = frcnn_config.first_stage_objectness_loss_weight
initial_crop_size = frcnn_config.initial_crop_size
maxpool_kernel_size = frcnn_config.maxpool_kernel_size
maxpool_stride = frcnn_config.maxpool_stride
second_stage_target_assigner = target_assigner.create_target_assigner(
'FasterRCNN',
'detection',
use_matmul_gather=frcnn_config.use_matmul_gather_in_matcher)
second_stage_box_predictor = box_predictor_builder.build(
hyperparams_builder.build,
frcnn_config.second_stage_box_predictor,
is_training=is_training,
num_classes=num_classes)
second_stage_batch_size = frcnn_config.second_stage_batch_size
second_stage_sampler = sampler.BalancedPositiveNegativeSampler(
positive_fraction=frcnn_config.second_stage_balance_fraction,
is_static=(frcnn_config.use_static_balanced_label_sampler and
use_static_shapes))
(second_stage_non_max_suppression_fn, second_stage_score_conversion_fn
) = post_processing_builder.build(frcnn_config.second_stage_post_processing)
second_stage_localization_loss_weight = (
frcnn_config.second_stage_localization_loss_weight)
second_stage_classification_loss = (
losses_builder.build_faster_rcnn_classification_loss(
frcnn_config.second_stage_classification_loss))
second_stage_classification_loss_weight = (
frcnn_config.second_stage_classification_loss_weight)
second_stage_mask_prediction_loss_weight = (
frcnn_config.second_stage_mask_prediction_loss_weight)
hard_example_miner = None
if frcnn_config.HasField('hard_example_miner'):
hard_example_miner = losses_builder.build_hard_example_miner(
frcnn_config.hard_example_miner,
second_stage_classification_loss_weight,
second_stage_localization_loss_weight)
crop_and_resize_fn = (
ops.matmul_crop_and_resize if frcnn_config.use_matmul_crop_and_resize
else ops.native_crop_and_resize)
clip_anchors_to_image = (
frcnn_config.clip_anchors_to_image)
common_kwargs = {
'is_training': is_training,
'num_classes': num_classes,
'image_resizer_fn': image_resizer_fn,
'feature_extractor': feature_extractor,
'number_of_stages': number_of_stages,
'first_stage_anchor_generator': first_stage_anchor_generator,
'first_stage_target_assigner': first_stage_target_assigner,
'first_stage_atrous_rate': first_stage_atrous_rate,
'first_stage_box_predictor_arg_scope_fn':
first_stage_box_predictor_arg_scope_fn,
'first_stage_box_predictor_kernel_size':
first_stage_box_predictor_kernel_size,
'first_stage_box_predictor_depth': first_stage_box_predictor_depth,
'first_stage_minibatch_size': first_stage_minibatch_size,
'first_stage_sampler': first_stage_sampler,
'first_stage_non_max_suppression_fn': first_stage_non_max_suppression_fn,
'first_stage_max_proposals': first_stage_max_proposals,
'first_stage_localization_loss_weight': first_stage_loc_loss_weight,
'first_stage_objectness_loss_weight': first_stage_obj_loss_weight,
'second_stage_target_assigner': second_stage_target_assigner,
'second_stage_batch_size': second_stage_batch_size,
'second_stage_sampler': second_stage_sampler,
'second_stage_non_max_suppression_fn':
second_stage_non_max_suppression_fn,
'second_stage_score_conversion_fn': second_stage_score_conversion_fn,
'second_stage_localization_loss_weight':
second_stage_localization_loss_weight,
'second_stage_classification_loss':
second_stage_classification_loss,
'second_stage_classification_loss_weight':
second_stage_classification_loss_weight,
'hard_example_miner': hard_example_miner,
'add_summaries': add_summaries,
'crop_and_resize_fn': crop_and_resize_fn,
'clip_anchors_to_image': clip_anchors_to_image,
'use_static_shapes': use_static_shapes,
'resize_masks': frcnn_config.resize_masks
}
filter_fn_arg = kwargs.get('filter_fn_arg')
if filter_fn_arg:
filter_fn = functools.partial(filter_bbox, **filter_fn_arg)
common_kwargs['filter_fn'] = filter_fn
rpn_type = kwargs.get('rpn_type')
if rpn_type:
common_kwargs['rpn_type'] = rpn_type
replace_rpn_arg = kwargs.get('replace_rpn_arg')
if replace_rpn_arg:
common_kwargs['replace_rpn_arg'] = replace_rpn_arg
if isinstance(second_stage_box_predictor,
rfcn_box_predictor.RfcnBoxPredictor):
return rfcn_meta_arch.RFCNMetaArch(
second_stage_rfcn_box_predictor=second_stage_box_predictor,
**common_kwargs)
else:
return faster_rcnn_meta_arch.FasterRCNNMetaArch(
initial_crop_size=initial_crop_size,
maxpool_kernel_size=maxpool_kernel_size,
maxpool_stride=maxpool_stride,
second_stage_mask_rcnn_box_predictor=second_stage_box_predictor,
second_stage_mask_prediction_loss_weight=(
second_stage_mask_prediction_loss_weight),
**common_kwargs)
"""
first_stage_nms_score_threshold = frcnn_config.first_stage_nms_score_threshold #they gave it as zero
first_stage_nms_iou_threshold = frcnn_config.first_stage_nms_iou_threshold #gave it 0.7
first_stage_max_proposals = frcnn_config.first_stage_max_proposals #how many proposals in the first stage
first_stage_loc_loss_weight = (
frcnn_config.first_stage_localization_loss_weight) #This is the weight param related to regression loss in the rpn loss function
first_stage_obj_loss_weight = frcnn_config.first_stage_objectness_loss_weight #same
initial_crop_size = frcnn_config.initial_crop_size #crop size ?? not sure I think the feature map size
maxpool_kernel_size = frcnn_config.maxpool_kernel_size #ppoling kernal not sure
maxpool_stride = frcnn_config.maxpool_stride # not sure
second_stage_box_predictor = box_predictor_builder.build( #This will predict the boxes
hyperparams_builder.build, #argoarse function retun inoder to create the box predictort (This is after the prediction frm rpn)
frcnn_config.second_stage_box_predictor, #variables from the config file
is_training=is_training,
num_classes=num_classes)
second_stage_batch_size = frcnn_config.second_stage_batch_size #not given
second_stage_balance_fraction = frcnn_config.second_stage_balance_fraction #not given
#here this one will output the
(second_stage_non_max_suppression_fn, second_stage_score_conversion_fn #this is for post processing of real predicted bpces and stuff
) = post_processing_builder.build(frcnn_config.second_stage_post_processing) #output two funtions
second_stage_localization_loss_weight = ( #again for the loss function
frcnn_config.second_stage_localization_loss_weight)
second_stage_classification_loss_weight = ( #again for the joint loss function
frcnn_config.second_stage_classification_loss_weight)
def _build_ssd_model(ssd_config, is_training, add_summaries):
"""Builds an SSD detection model based on the model config.
Args:
ssd_config: A ssd.proto object containing the config for the desired
SSDMetaArch.
is_training: True if this model is being built for training purposes.
add_summaries: Whether to add tf summaries in the model.
Returns:
SSDMetaArch based on the config.
Raises:
ValueError: If ssd_config.type is not recognized (i.e. not registered in
model_class_map).
"""
num_classes = ssd_config.num_classes
# Feature extractor
feature_extractor = _build_ssd_feature_extractor(
ssd_config.feature_extractor, is_training)
box_coder = box_coder_builder.build(ssd_config.box_coder)
matcher = matcher_builder.build(ssd_config.matcher)
region_similarity_calculator = sim_calc.build(
ssd_config.similarity_calculator)
encode_background_as_zeros = ssd_config.encode_background_as_zeros
negative_class_weight = ssd_config.negative_class_weight
ssd_box_predictor = box_predictor_builder.build(hyperparams_builder.build,
ssd_config.box_predictor,
is_training, num_classes)
anchor_generator = anchor_generator_builder.build(
ssd_config.anchor_generator)
image_resizer_fn = image_resizer_builder.build(ssd_config.image_resizer)
non_max_suppression_fn, score_conversion_fn = post_processing_builder.build(
ssd_config.post_processing)
(classification_loss, localization_loss, classification_weight,
localization_weight,
hard_example_miner) = losses_builder.build(ssd_config.loss)
normalize_loss_by_num_matches = ssd_config.normalize_loss_by_num_matches
normalize_loc_loss_by_codesize = ssd_config.normalize_loc_loss_by_codesize
return ssd_meta_arch.SSDMetaArch(
is_training,
anchor_generator,
ssd_box_predictor,
box_coder,
feature_extractor,
matcher,
region_similarity_calculator,
encode_background_as_zeros,
negative_class_weight,
image_resizer_fn,
non_max_suppression_fn,
score_conversion_fn,
classification_loss,
localization_loss,
classification_weight,
localization_weight,
normalize_loss_by_num_matches,
hard_example_miner,
add_summaries=add_summaries,
normalize_loc_loss_by_codesize=normalize_loc_loss_by_codesize)
def _build_ssd_model(ssd_config,
is_training,
add_summaries,
add_background_class=True):
"""Builds an SSD detection model based on the model config.
Args:
ssd_config: A ssd.proto object containing the config for the desired
SSDMetaArch.
is_training: True if this model is being built for training purposes.
add_summaries: Whether to add tf summaries in the model.
add_background_class: Whether to add an implicit background class to one-hot
encodings of groundtruth labels. Set to false if using groundtruth labels
with an explicit background class or using multiclass scores instead of
truth in the case of distillation.
Returns:
SSDMetaArch based on the config.
Raises:
ValueError: If ssd_config.type is not recognized (i.e. not registered in
model_class_map).
"""
num_classes = ssd_config.num_classes
# Feature extractor
feature_extractor = _build_ssd_feature_extractor(
feature_extractor_config=ssd_config.feature_extractor,
is_training=is_training)
box_coder = box_coder_builder.build(ssd_config.box_coder)
matcher = matcher_builder.build(ssd_config.matcher)
region_similarity_calculator = sim_calc.build(
ssd_config.similarity_calculator)
encode_background_as_zeros = ssd_config.encode_background_as_zeros
negative_class_weight = ssd_config.negative_class_weight
ssd_box_predictor = box_predictor_builder.build(hyperparams_builder.build,
ssd_config.box_predictor,
is_training, num_classes)
anchor_generator = anchor_generator_builder.build(
ssd_config.anchor_generator)
image_resizer_fn = image_resizer_builder.build(ssd_config.image_resizer)
non_max_suppression_fn, score_conversion_fn = post_processing_builder.build(
ssd_config.post_processing)
(classification_loss, localization_loss, classification_weight,
localization_weight, hard_example_miner,
random_example_sampler) = losses_builder.build(ssd_config.loss)
normalize_loss_by_num_matches = ssd_config.normalize_loss_by_num_matches
normalize_loc_loss_by_codesize = ssd_config.normalize_loc_loss_by_codesize
weight_regression_loss_by_score = (
ssd_config.weight_regression_loss_by_score)
target_assigner_instance = target_assigner.TargetAssigner(
region_similarity_calculator,
matcher,
box_coder,
negative_class_weight=negative_class_weight,
weight_regression_loss_by_score=weight_regression_loss_by_score)
expected_classification_loss_under_sampling = None
if ssd_config.use_expected_classification_loss_under_sampling:
expected_classification_loss_under_sampling = functools.partial(
ops.expected_classification_loss_under_sampling,
minimum_negative_sampling=ssd_config.minimum_negative_sampling,
desired_negative_sampling_ratio=ssd_config.
desired_negative_sampling_ratio)
ssd_meta_arch_fn = ssd_meta_arch.SSDMetaArch
# BEGIN GOOGLE-INTERNAL
# TODO(lzc): move ssd_mask_meta_arch to third party when it has decent
# performance relative to a comparable Mask R-CNN model (b/112561592).
predictor_config = ssd_config.box_predictor
predict_instance_masks = False
if predictor_config.WhichOneof(
'box_predictor_oneof') == 'convolutional_box_predictor':
predict_instance_masks = (
predictor_config.convolutional_box_predictor.HasField('mask_head'))
elif predictor_config.WhichOneof(
'box_predictor_oneof'
) == 'weight_shared_convolutional_box_predictor':
predict_instance_masks = (
predictor_config.weight_shared_convolutional_box_predictor.
HasField('mask_head'))
if predict_instance_masks:
ssd_meta_arch_fn = ssd_mask_meta_arch.SSDMaskMetaArch
# END GOOGLE-INTERNAL
return ssd_meta_arch_fn(
is_training=is_training,
anchor_generator=anchor_generator,
box_predictor=ssd_box_predictor,
box_coder=box_coder,
feature_extractor=feature_extractor,
encode_background_as_zeros=encode_background_as_zeros,
image_resizer_fn=image_resizer_fn,
non_max_suppression_fn=non_max_suppression_fn,
score_conversion_fn=score_conversion_fn,
classification_loss=classification_loss,
localization_loss=localization_loss,
classification_loss_weight=classification_weight,
localization_loss_weight=localization_weight,
normalize_loss_by_num_matches=normalize_loss_by_num_matches,
hard_example_miner=hard_example_miner,
target_assigner_instance=target_assigner_instance,
add_summaries=add_summaries,
normalize_loc_loss_by_codesize=normalize_loc_loss_by_codesize,
freeze_batchnorm=ssd_config.freeze_batchnorm,
inplace_batchnorm_update=ssd_config.inplace_batchnorm_update,
add_background_class=add_background_class,
random_example_sampler=random_example_sampler,
expected_classification_loss_under_sampling=
expected_classification_loss_under_sampling)
def _build_faster_rcnn_model(frcnn_config,
is_training,
add_summaries,
meta_architecture='faster_rcnn'):
"""Builds a Faster R-CNN or R-FCN detection model based on the model config.
Builds R-FCN model if the second_stage_box_predictor in the config is of type
`rfcn_box_predictor` else builds a Faster R-CNN model.
Args:
frcnn_config: A faster_rcnn.proto object containing the config for the
desired FasterRCNNMetaArch or RFCNMetaArch.
is_training: True if this model is being built for training purposes.
add_summaries: Whether to add tf summaries in the model.
Returns:
FasterRCNNMetaArch based on the config.
Raises:
ValueError: If frcnn_config.type is not recognized (i.e. not registered in
model_class_map).
"""
num_classes = frcnn_config.num_classes
image_resizer_fn = image_resizer_builder.build(frcnn_config.image_resizer)
feature_extractor = _build_faster_rcnn_feature_extractor(
frcnn_config.feature_extractor, is_training,
frcnn_config.inplace_batchnorm_update)
number_of_stages = frcnn_config.number_of_stages
first_stage_anchor_generator = anchor_generator_builder.build(
frcnn_config.first_stage_anchor_generator)
first_stage_target_assigner = target_assigner.create_target_assigner(
'FasterRCNN',
'proposal',
use_matmul_gather=frcnn_config.use_matmul_gather_in_matcher)
first_stage_atrous_rate = frcnn_config.first_stage_atrous_rate
first_stage_box_predictor_arg_scope_fn = hyperparams_builder.build(
frcnn_config.first_stage_box_predictor_conv_hyperparams, is_training)
first_stage_box_predictor_kernel_size = (
frcnn_config.first_stage_box_predictor_kernel_size)
first_stage_box_predictor_depth = frcnn_config.first_stage_box_predictor_depth
first_stage_minibatch_size = frcnn_config.first_stage_minibatch_size
# TODO(bhattad): When eval is supported using static shapes, add separate
# use_static_shapes_for_trainig and use_static_shapes_for_evaluation.
use_static_shapes = frcnn_config.use_static_shapes and is_training
first_stage_sampler = sampler.BalancedPositiveNegativeSampler(
positive_fraction=frcnn_config.first_stage_positive_balance_fraction,
is_static=frcnn_config.use_static_balanced_label_sampler
and is_training)
first_stage_max_proposals = frcnn_config.first_stage_max_proposals
first_stage_proposals_path = frcnn_config.first_stage_proposals_path
if (frcnn_config.first_stage_nms_iou_threshold < 0
or frcnn_config.first_stage_nms_iou_threshold > 1.0):
raise ValueError('iou_threshold not in [0, 1.0].')
if (is_training and
frcnn_config.second_stage_batch_size > first_stage_max_proposals):
raise ValueError('second_stage_batch_size should be no greater than '
'first_stage_max_proposals.')
first_stage_non_max_suppression_fn = functools.partial(
post_processing.batch_multiclass_non_max_suppression,
score_thresh=frcnn_config.first_stage_nms_score_threshold,
iou_thresh=frcnn_config.first_stage_nms_iou_threshold,
max_size_per_class=frcnn_config.first_stage_max_proposals,
max_total_size=frcnn_config.first_stage_max_proposals,
use_static_shapes=use_static_shapes and is_training)
first_stage_loc_loss_weight = (
frcnn_config.first_stage_localization_loss_weight)
first_stage_obj_loss_weight = frcnn_config.first_stage_objectness_loss_weight
initial_crop_size = frcnn_config.initial_crop_size
maxpool_kernel_size = frcnn_config.maxpool_kernel_size
maxpool_stride = frcnn_config.maxpool_stride
second_stage_target_assigner = target_assigner.create_target_assigner(
'FasterRCNN',
'detection',
use_matmul_gather=frcnn_config.use_matmul_gather_in_matcher,
iou_threshold=frcnn_config.second_stage_target_iou_threshold)
second_stage_box_predictor = box_predictor_builder.build(
hyperparams_builder.build,
frcnn_config.second_stage_box_predictor,
is_training=is_training,
num_classes=num_classes)
second_stage_batch_size = frcnn_config.second_stage_batch_size
second_stage_sampler = sampler.BalancedPositiveNegativeSampler(
positive_fraction=frcnn_config.second_stage_balance_fraction,
is_static=frcnn_config.use_static_balanced_label_sampler
and is_training)
(second_stage_non_max_suppression_fn,
second_stage_score_conversion_fn) = post_processing_builder.build(
frcnn_config.second_stage_post_processing)
second_stage_localization_loss_weight = (
frcnn_config.second_stage_localization_loss_weight)
second_stage_classification_loss = (
losses_builder.build_faster_rcnn_classification_loss(
frcnn_config.second_stage_classification_loss))
second_stage_classification_loss_weight = (
frcnn_config.second_stage_classification_loss_weight)
second_stage_mask_prediction_loss_weight = (
frcnn_config.second_stage_mask_prediction_loss_weight)
hard_example_miner = None
if frcnn_config.HasField('hard_example_miner'):
hard_example_miner = losses_builder.build_hard_example_miner(
frcnn_config.hard_example_miner,
second_stage_classification_loss_weight,
second_stage_localization_loss_weight)
crop_and_resize_fn = (ops.matmul_crop_and_resize
if frcnn_config.use_matmul_crop_and_resize else
ops.native_crop_and_resize)
clip_anchors_to_image = (frcnn_config.clip_anchors_to_image)
common_kwargs = {
'is_training': is_training,
'num_classes': num_classes,
'image_resizer_fn': image_resizer_fn,
'feature_extractor': feature_extractor,
'number_of_stages': number_of_stages,
'first_stage_anchor_generator': first_stage_anchor_generator,
'first_stage_target_assigner': first_stage_target_assigner,
'first_stage_atrous_rate': first_stage_atrous_rate,
'first_stage_box_predictor_arg_scope_fn':
first_stage_box_predictor_arg_scope_fn,
'first_stage_box_predictor_kernel_size':
first_stage_box_predictor_kernel_size,
'first_stage_box_predictor_depth': first_stage_box_predictor_depth,
'first_stage_minibatch_size': first_stage_minibatch_size,
'first_stage_sampler': first_stage_sampler,
'first_stage_non_max_suppression_fn':
first_stage_non_max_suppression_fn,
'first_stage_max_proposals': first_stage_max_proposals,
'first_stage_localization_loss_weight': first_stage_loc_loss_weight,
'first_stage_objectness_loss_weight': first_stage_obj_loss_weight,
'second_stage_target_assigner': second_stage_target_assigner,
'second_stage_batch_size': second_stage_batch_size,
'second_stage_sampler': second_stage_sampler,
'second_stage_non_max_suppression_fn':
second_stage_non_max_suppression_fn,
'second_stage_score_conversion_fn': second_stage_score_conversion_fn,
'second_stage_localization_loss_weight':
second_stage_localization_loss_weight,
'second_stage_classification_loss': second_stage_classification_loss,
'second_stage_classification_loss_weight':
second_stage_classification_loss_weight,
'hard_example_miner': hard_example_miner,
'add_summaries': add_summaries,
'crop_and_resize_fn': crop_and_resize_fn,
'clip_anchors_to_image': clip_anchors_to_image,
'use_static_shapes': use_static_shapes,
'resize_masks': frcnn_config.resize_masks
}
if isinstance(second_stage_box_predictor,
rfcn_box_predictor.RfcnBoxPredictor):
return rfcn_meta_arch.RFCNMetaArch(
second_stage_rfcn_box_predictor=second_stage_box_predictor,
**common_kwargs)
elif meta_architecture == 'faster_rcnn':
return faster_rcnn_meta_arch.FasterRCNNMetaArch(
initial_crop_size=initial_crop_size,
maxpool_kernel_size=maxpool_kernel_size,
maxpool_stride=maxpool_stride,
second_stage_mask_rcnn_box_predictor=second_stage_box_predictor,
second_stage_mask_prediction_loss_weight=(
second_stage_mask_prediction_loss_weight),
**common_kwargs)
elif meta_architecture == 'faster_rcnn_override_RPN':
return faster_rcnn_meta_arch_override_RPN.FasterRCNNMetaArchOverrideRPN(
initial_crop_size=initial_crop_size,
maxpool_kernel_size=maxpool_kernel_size,
maxpool_stride=maxpool_stride,
first_stage_proposals_path=first_stage_proposals_path,
second_stage_mask_rcnn_box_predictor=second_stage_box_predictor,
second_stage_mask_prediction_loss_weight=(
second_stage_mask_prediction_loss_weight),
**common_kwargs)
elif meta_architecture == 'faster_rcnn_rpn_blend':
common_kwargs['use_matmul_crop_and_resize'] = False
common_kwargs[
'first_stage_nms_iou_threshold'] = frcnn_config.first_stage_nms_iou_threshold
common_kwargs[
'first_stage_nms_score_threshold'] = frcnn_config.first_stage_nms_score_threshold
common_kwargs.pop('crop_and_resize_fn')
common_kwargs.pop('first_stage_non_max_suppression_fn')
common_kwargs.pop('resize_masks')
common_kwargs.pop('use_static_shapes')
return faster_rcnn_meta_arch_rpn_blend.FasterRCNNMetaArchRPNBlend(
initial_crop_size=initial_crop_size,
maxpool_kernel_size=maxpool_kernel_size,
maxpool_stride=maxpool_stride,
first_stage_proposals_path=first_stage_proposals_path,
second_stage_mask_rcnn_box_predictor=second_stage_box_predictor,
second_stage_mask_prediction_loss_weight=(
second_stage_mask_prediction_loss_weight),
**common_kwargs)
def _build_ssd_model(ssd_config, is_training, add_summaries,
add_background_class=True):
"""Builds an SSD detection model based on the model config.
Args:
ssd_config: A ssd.proto object containing the config for the desired
SSDMetaArch.
is_training: True if this model is being built for training purposes.
add_summaries: Whether to add tf summaries in the model.
add_background_class: Whether to add an implicit background class to one-hot
encodings of groundtruth labels. Set to false if using groundtruth labels
with an explicit background class or using multiclass scores instead of
truth in the case of distillation.
Returns:
SSDMetaArch based on the config.
Raises:
ValueError: If ssd_config.type is not recognized (i.e. not registered in
model_class_map).
"""
num_classes = ssd_config.num_classes
# Feature extractor
feature_extractor = _build_ssd_feature_extractor(
feature_extractor_config=ssd_config.feature_extractor,
is_training=is_training)
box_coder = box_coder_builder.build(ssd_config.box_coder)
matcher = matcher_builder.build(ssd_config.matcher)
region_similarity_calculator = sim_calc.build(
ssd_config.similarity_calculator)
encode_background_as_zeros = ssd_config.encode_background_as_zeros
negative_class_weight = ssd_config.negative_class_weight
ssd_box_predictor = box_predictor_builder.build(hyperparams_builder.build,
ssd_config.box_predictor,
is_training, num_classes)
anchor_generator = anchor_generator_builder.build(
ssd_config.anchor_generator)
image_resizer_fn = image_resizer_builder.build(ssd_config.image_resizer)
non_max_suppression_fn, score_conversion_fn = post_processing_builder.build(
ssd_config.post_processing)
(classification_loss, localization_loss, classification_weight,
localization_weight,
hard_example_miner) = losses_builder.build(ssd_config.loss)
normalize_loss_by_num_matches = ssd_config.normalize_loss_by_num_matches
normalize_loc_loss_by_codesize = ssd_config.normalize_loc_loss_by_codesize
return ssd_meta_arch.SSDMetaArch(
is_training,
anchor_generator,
ssd_box_predictor,
box_coder,
feature_extractor,
matcher,
region_similarity_calculator,
encode_background_as_zeros,
negative_class_weight,
image_resizer_fn,
non_max_suppression_fn,
score_conversion_fn,
classification_loss,
localization_loss,
classification_weight,
localization_weight,
normalize_loss_by_num_matches,
hard_example_miner,
add_summaries=add_summaries,
normalize_loc_loss_by_codesize=normalize_loc_loss_by_codesize,
freeze_batchnorm=ssd_config.freeze_batchnorm,
inplace_batchnorm_update=ssd_config.inplace_batchnorm_update,
add_background_class=add_background_class)
def _build_lstm_model(ssd_config, lstm_config, is_training):
"""Builds an LSTM detection model based on the model config.
Args:
ssd_config: A ssd.proto object containing the config for the desired
LSTMSSDMetaArch.
lstm_config: LstmModel config proto that specifies LSTM train/eval configs.
is_training: True if this model is being built for training purposes.
Returns:
LSTMSSDMetaArch based on the config.
Raises:
ValueError: If ssd_config.type is not recognized (i.e. not registered in
model_class_map), or if lstm_config.interleave_strategy is not recognized.
ValueError: If unroll_length is not specified in the config file.
"""
feature_extractor = _build_lstm_feature_extractor(
ssd_config.feature_extractor, is_training, lstm_config)
box_coder = box_coder_builder.build(ssd_config.box_coder)
matcher = matcher_builder.build(ssd_config.matcher)
region_similarity_calculator = sim_calc.build(
ssd_config.similarity_calculator)
num_classes = ssd_config.num_classes
ssd_box_predictor = box_predictor_builder.build(hyperparams_builder.build,
ssd_config.box_predictor,
is_training, num_classes)
anchor_generator = anchor_generator_builder.build(ssd_config.anchor_generator)
image_resizer_fn = image_resizer_builder.build(ssd_config.image_resizer)
non_max_suppression_fn, score_conversion_fn = post_processing_builder.build(
ssd_config.post_processing)
(classification_loss, localization_loss, classification_weight,
localization_weight, miner, _, _) = losses_builder.build(ssd_config.loss)
normalize_loss_by_num_matches = ssd_config.normalize_loss_by_num_matches
encode_background_as_zeros = ssd_config.encode_background_as_zeros
negative_class_weight = ssd_config.negative_class_weight
# Extra configs for lstm unroll length.
unroll_length = None
if 'lstm' in ssd_config.feature_extractor.type:
if is_training:
unroll_length = lstm_config.train_unroll_length
else:
unroll_length = lstm_config.eval_unroll_length
if unroll_length is None:
raise ValueError('No unroll length found in the config file')
target_assigner_instance = target_assigner.TargetAssigner(
region_similarity_calculator,
matcher,
box_coder,
negative_class_weight=negative_class_weight)
lstm_model = lstm_ssd_meta_arch.LSTMSSDMetaArch(
is_training=is_training,
anchor_generator=anchor_generator,
box_predictor=ssd_box_predictor,
box_coder=box_coder,
feature_extractor=feature_extractor,
encode_background_as_zeros=encode_background_as_zeros,
image_resizer_fn=image_resizer_fn,
non_max_suppression_fn=non_max_suppression_fn,
score_conversion_fn=score_conversion_fn,
classification_loss=classification_loss,
localization_loss=localization_loss,
classification_loss_weight=classification_weight,
localization_loss_weight=localization_weight,
normalize_loss_by_num_matches=normalize_loss_by_num_matches,
hard_example_miner=miner,
unroll_length=unroll_length,
target_assigner_instance=target_assigner_instance)
return lstm_model
def _build_faster_rcnn_model(frcnn_config, is_training, mtl=None):
"""Builds a Faster R-CNN or R-FCN detection model based on the model config.
Builds R-FCN model if the second_stage_box_predictor in the config is of type
`rfcn_box_predictor` else builds a Faster R-CNN model.
Args:
frcnn_config: A faster_rcnn.proto object containing the config for the
desired FasterRCNNMetaArch or RFCNMetaArch.
is_training: True if this model is being built for training purposes.
Returns:
FasterRCNNMetaArch based on the config.
Raises:
ValueError: If frcnn_config.type is not recognized (i.e. not registered in
model_class_map).
"""
num_classes = frcnn_config.num_classes
image_resizer_fn = image_resizer_builder.build(frcnn_config.image_resizer)
feature_extractor_kwargs = {}
feature_extractor_kwargs[
'freeze_layer'] = frcnn_config.feature_extractor.freeze_layer
feature_extractor_kwargs[
'batch_norm_trainable'] = frcnn_config.feature_extractor.batch_norm_trainable
if frcnn_config.feature_extractor.HasField('weight_decay'):
feature_extractor_kwargs['weight_decay'] = \
frcnn_config.feature_extractor.weight_decay
feature_extractor = _build_faster_rcnn_feature_extractor(
frcnn_config.feature_extractor,
is_training and frcnn_config.feature_extractor.trainable,
reuse_weights=tf.AUTO_REUSE,
**feature_extractor_kwargs)
first_stage_only = frcnn_config.first_stage_only
first_stage_anchor_generator = anchor_generator_builder.build(
frcnn_config.first_stage_anchor_generator)
first_stage_clip_window = frcnn_config.first_stage_clip_window
first_stage_atrous_rate = frcnn_config.first_stage_atrous_rate
first_stage_box_predictor_trainable = \
frcnn_config.first_stage_box_predictor_trainable
first_stage_box_predictor_arg_scope = hyperparams_builder.build(
frcnn_config.first_stage_box_predictor_conv_hyperparams, is_training)
first_stage_box_predictor_kernel_size = (
frcnn_config.first_stage_box_predictor_kernel_size)
first_stage_box_predictor_depth = frcnn_config.first_stage_box_predictor_depth
first_stage_minibatch_size = frcnn_config.first_stage_minibatch_size
first_stage_positive_balance_fraction = (
frcnn_config.first_stage_positive_balance_fraction)
first_stage_nms_score_threshold = frcnn_config.first_stage_nms_score_threshold
first_stage_nms_iou_threshold = frcnn_config.first_stage_nms_iou_threshold
first_stage_max_proposals = frcnn_config.first_stage_max_proposals
first_stage_loc_loss_weight = (
frcnn_config.first_stage_localization_loss_weight)
first_stage_obj_loss_weight = frcnn_config.first_stage_objectness_loss_weight
initial_crop_size = frcnn_config.initial_crop_size
maxpool_kernel_size = frcnn_config.maxpool_kernel_size
maxpool_stride = frcnn_config.maxpool_stride
second_stage_box_predictor = box_predictor_builder.build(
hyperparams_builder.build,
frcnn_config.second_stage_box_predictor,
is_training=is_training
and frcnn_config.second_stage_box_predictor.trainable,
num_classes=num_classes,
reuse_weights=tf.AUTO_REUSE)
second_stage_batch_size = frcnn_config.second_stage_batch_size
second_stage_balance_fraction = frcnn_config.second_stage_balance_fraction
(second_stage_non_max_suppression_fn,
second_stage_score_conversion_fn) = post_processing_builder.build(
frcnn_config.second_stage_post_processing)
second_stage_localization_loss_weight = (
frcnn_config.second_stage_localization_loss_weight)
second_stage_classification_loss_weight = (
frcnn_config.second_stage_classification_loss_weight)
if mtl.window:
window_box_predictor = box_predictor_builder.build(
hyperparams_builder.build,
mtl.window_box_predictor,
is_training=is_training and mtl.window_box_predictor.trainable,
num_classes=num_classes + 1,
reuse_weights=tf.AUTO_REUSE)
else:
window_box_predictor = second_stage_box_predictor
if mtl.closeness:
closeness_box_predictor = box_predictor_builder.build(
hyperparams_builder.build,
mtl.closeness_box_predictor,
is_training=is_training and mtl.closeness_box_predictor.trainable,
num_classes=num_classes + 1,
reuse_weights=tf.AUTO_REUSE)
else:
closeness_box_predictor = second_stage_box_predictor
if mtl.edgemask:
edgemask_predictor = mask_predictor_builder.build(
hyperparams_builder.build,
mtl.edgemask_predictor,
is_training=is_training and mtl.edgemask_predictor.trainable,
num_classes=2,
reuse_weights=tf.AUTO_REUSE,
channels=1)
else:
edgemask_predictor = None
mtl_refiner_arg_scope = None
if mtl.refine:
mtl_refiner_arg_scope = hyperparams_builder.build(
mtl.refiner_fc_hyperparams, is_training)
hard_example_miner = None
if frcnn_config.HasField('hard_example_miner'):
hard_example_miner = losses_builder.build_hard_example_miner(
frcnn_config.hard_example_miner,
second_stage_classification_loss_weight,
second_stage_localization_loss_weight)
common_kwargs = {
'is_training': is_training,
'num_classes': num_classes,
'image_resizer_fn': image_resizer_fn,
'feature_extractor': feature_extractor,
'first_stage_only': first_stage_only,
'first_stage_anchor_generator': first_stage_anchor_generator,
'first_stage_clip_window': first_stage_clip_window,
'first_stage_atrous_rate': first_stage_atrous_rate,
'first_stage_box_predictor_trainable':
first_stage_box_predictor_trainable,
'first_stage_box_predictor_arg_scope':
first_stage_box_predictor_arg_scope,
'first_stage_box_predictor_kernel_size':
first_stage_box_predictor_kernel_size,
'first_stage_box_predictor_depth': first_stage_box_predictor_depth,
'first_stage_minibatch_size': first_stage_minibatch_size,
'first_stage_positive_balance_fraction':
first_stage_positive_balance_fraction,
'first_stage_nms_score_threshold': first_stage_nms_score_threshold,
'first_stage_nms_iou_threshold': first_stage_nms_iou_threshold,
'first_stage_max_proposals': first_stage_max_proposals,
'first_stage_localization_loss_weight': first_stage_loc_loss_weight,
'first_stage_objectness_loss_weight': first_stage_obj_loss_weight,
'second_stage_batch_size': second_stage_batch_size,
'second_stage_balance_fraction': second_stage_balance_fraction,
'second_stage_non_max_suppression_fn':
second_stage_non_max_suppression_fn,
'second_stage_score_conversion_fn': second_stage_score_conversion_fn,
'second_stage_localization_loss_weight':
second_stage_localization_loss_weight,
'second_stage_classification_loss_weight':
second_stage_classification_loss_weight,
'hard_example_miner': hard_example_miner,
'mtl': mtl,
'mtl_refiner_arg_scope': mtl_refiner_arg_scope,
'window_box_predictor': window_box_predictor,
'closeness_box_predictor': closeness_box_predictor,
'edgemask_predictor': edgemask_predictor
}
if isinstance(second_stage_box_predictor, box_predictor.RfcnBoxPredictor):
return rfcn_meta_arch.RFCNMetaArch(
second_stage_rfcn_box_predictor=second_stage_box_predictor,
**common_kwargs)
else:
return faster_rcnn_meta_arch.FasterRCNNMetaArch(
initial_crop_size=initial_crop_size,
maxpool_kernel_size=maxpool_kernel_size,
maxpool_stride=maxpool_stride,
second_stage_mask_rcnn_box_predictor=second_stage_box_predictor,
**common_kwargs)
def _build_lstm_model(ssd_config, lstm_config, is_training):
"""Builds an LSTM detection model based on the model config.
Args:
ssd_config: A ssd.proto object containing the config for the desired
LSTMMetaArch.
lstm_config: LstmModel config proto that specifies LSTM train/eval configs.
is_training: True if this model is being built for training purposes.
Returns:
LSTMMetaArch based on the config.
Raises:
ValueError: If ssd_config.type is not recognized (i.e. not registered in
model_class_map), or if lstm_config.interleave_strategy is not recognized.
ValueError: If unroll_length is not specified in the config file.
"""
feature_extractor = _build_lstm_feature_extractor(
ssd_config.feature_extractor, is_training, lstm_config.lstm_state_depth)
box_coder = box_coder_builder.build(ssd_config.box_coder)
matcher = matcher_builder.build(ssd_config.matcher)
region_similarity_calculator = sim_calc.build(
ssd_config.similarity_calculator)
num_classes = ssd_config.num_classes
ssd_box_predictor = box_predictor_builder.build(hyperparams_builder.build,
ssd_config.box_predictor,
is_training, num_classes)
anchor_generator = anchor_generator_builder.build(ssd_config.anchor_generator)
image_resizer_fn = image_resizer_builder.build(ssd_config.image_resizer)
non_max_suppression_fn, score_conversion_fn = post_processing_builder.build(
ssd_config.post_processing)
(classification_loss, localization_loss, classification_weight,
localization_weight, miner, _, _) = losses_builder.build(ssd_config.loss)
normalize_loss_by_num_matches = ssd_config.normalize_loss_by_num_matches
encode_background_as_zeros = ssd_config.encode_background_as_zeros
negative_class_weight = ssd_config.negative_class_weight
# Extra configs for lstm unroll length.
unroll_length = None
if 'lstm' in ssd_config.feature_extractor.type:
if is_training:
unroll_length = lstm_config.train_unroll_length
else:
unroll_length = lstm_config.eval_unroll_length
if unroll_length is None:
raise ValueError('No unroll length found in the config file')
target_assigner_instance = target_assigner.TargetAssigner(
region_similarity_calculator,
matcher,
box_coder,
negative_class_weight=negative_class_weight)
lstm_model = lstm_meta_arch.LSTMMetaArch(
is_training=is_training,
anchor_generator=anchor_generator,
box_predictor=ssd_box_predictor,
box_coder=box_coder,
feature_extractor=feature_extractor,
encode_background_as_zeros=encode_background_as_zeros,
image_resizer_fn=image_resizer_fn,
non_max_suppression_fn=non_max_suppression_fn,
score_conversion_fn=score_conversion_fn,
classification_loss=classification_loss,
localization_loss=localization_loss,
classification_loss_weight=classification_weight,
localization_loss_weight=localization_weight,
normalize_loss_by_num_matches=normalize_loss_by_num_matches,
hard_example_miner=miner,
unroll_length=unroll_length,
target_assigner_instance=target_assigner_instance)
return lstm_model
def _build_ssd_model(ssd_config, is_training, add_summaries):
num_classes = ssd_config.num_classes
# Feature extractor
feature_extractor = _build_ssd_feature_extractor(
feature_extractor_config=ssd_config.feature_extractor,
freeze_batchnorm=ssd_config.freeze_batchnorm,
is_training=is_training)
box_coder = box_coder_builder.build(ssd_config.box_coder)
matcher = matcher_builder.build(ssd_config.matcher)
region_similarity_calculator = sim_calc.build(
ssd_config.similarity_calculator)
encode_background_as_zeros = ssd_config.encode_background_as_zeros
negative_class_weight = ssd_config.negative_class_weight
anchor_generator = anchor_generator_builder.build(
ssd_config.anchor_generator)
if feature_extractor.is_keras_model:
ssd_box_predictor = box_predictor_builder.build_keras(
hyperparams_fn=hyperparams_builder.KerasLayerHyperparams,
freeze_batchnorm=ssd_config.freeze_batchnorm,
inplace_batchnorm_update=False,
num_predictions_per_location_list=anchor_generator
.num_anchors_per_location(),
box_predictor_config=ssd_config.box_predictor,
is_training=is_training,
num_classes=num_classes,
add_background_class=ssd_config.add_background_class)
else:
ssd_box_predictor = box_predictor_builder.build(
hyperparams_builder.build, ssd_config.box_predictor, is_training,
num_classes, ssd_config.add_background_class)
image_resizer_fn = image_resizer_builder.build(ssd_config.image_resizer)
non_max_suppression_fn, score_conversion_fn = post_processing_builder.build(
ssd_config.post_processing)
(classification_loss, localization_loss, classification_weight,
localization_weight, hard_example_miner, random_example_sampler,
expected_loss_weights_fn) = losses_builder.build(ssd_config.loss)
normalize_loss_by_num_matches = ssd_config.normalize_loss_by_num_matches
normalize_loc_loss_by_codesize = ssd_config.normalize_loc_loss_by_codesize
equalization_loss_config = ops.EqualizationLossConfig(
weight=ssd_config.loss.equalization_loss.weight,
exclude_prefixes=ssd_config.loss.equalization_loss.exclude_prefixes)
target_assigner_instance = target_assigner.TargetAssigner(
region_similarity_calculator,
matcher,
box_coder,
negative_class_weight=negative_class_weight)
ssd_meta_arch_fn = ssd_meta_arch.SSDMetaArch
kwargs = {}
return ssd_meta_arch_fn(
is_training=is_training,
anchor_generator=anchor_generator,
box_predictor=ssd_box_predictor,
box_coder=box_coder,
feature_extractor=feature_extractor,
encode_background_as_zeros=encode_background_as_zeros,
image_resizer_fn=image_resizer_fn,
non_max_suppression_fn=non_max_suppression_fn,
score_conversion_fn=score_conversion_fn,
classification_loss=classification_loss,
localization_loss=localization_loss,
classification_loss_weight=classification_weight,
localization_loss_weight=localization_weight,
normalize_loss_by_num_matches=normalize_loss_by_num_matches,
hard_example_miner=hard_example_miner,
target_assigner_instance=target_assigner_instance,
add_summaries=add_summaries,
normalize_loc_loss_by_codesize=normalize_loc_loss_by_codesize,
freeze_batchnorm=ssd_config.freeze_batchnorm,
inplace_batchnorm_update=ssd_config.inplace_batchnorm_update,
add_background_class=ssd_config.add_background_class,
explicit_background_class=ssd_config.explicit_background_class,
random_example_sampler=random_example_sampler,
expected_loss_weights_fn=expected_loss_weights_fn,
use_confidences_as_targets=ssd_config.use_confidences_as_targets,
implicit_example_weight=ssd_config.implicit_example_weight,
equalization_loss_config=equalization_loss_config,
**kwargs)
def _build_faster_rcnn_model(frcnn_config, is_training, add_summaries):
"""Builds a Faster R-CNN or R-FCN detection model based on the model config.
Builds R-FCN model if the second_stage_box_predictor in the config is of type
`rfcn_box_predictor` else builds a Faster R-CNN model.
Args:
frcnn_config: A faster_rcnn.proto object containing the config for the
desired FasterRCNNMetaArch or RFCNMetaArch.
is_training: True if this model is being built for training purposes.
add_summaries: Whether to add tf summaries in the model.
Returns:
FasterRCNNMetaArch based on the config.
Raises:
ValueError: If frcnn_config.type is not recognized (i.e. not registered in
model_class_map).
"""
num_classes = frcnn_config.num_classes
image_resizer_fn = image_resizer_builder.build(frcnn_config.image_resizer)
feature_extractor = _build_faster_rcnn_feature_extractor(
frcnn_config.feature_extractor, is_training)
number_of_stages = frcnn_config.number_of_stages
first_stage_anchor_generator = anchor_generator_builder.build(
frcnn_config.first_stage_anchor_generator)
first_stage_atrous_rate = frcnn_config.first_stage_atrous_rate
first_stage_box_predictor_arg_scope = hyperparams_builder.build(
frcnn_config.first_stage_box_predictor_conv_hyperparams, is_training)
first_stage_box_predictor_kernel_size = (
frcnn_config.first_stage_box_predictor_kernel_size)
first_stage_box_predictor_depth = frcnn_config.first_stage_box_predictor_depth
first_stage_minibatch_size = frcnn_config.first_stage_minibatch_size
first_stage_positive_balance_fraction = (
frcnn_config.first_stage_positive_balance_fraction)
first_stage_nms_score_threshold = frcnn_config.first_stage_nms_score_threshold
first_stage_nms_iou_threshold = frcnn_config.first_stage_nms_iou_threshold
first_stage_max_proposals = frcnn_config.first_stage_max_proposals
first_stage_loc_loss_weight = (
frcnn_config.first_stage_localization_loss_weight)
first_stage_obj_loss_weight = frcnn_config.first_stage_objectness_loss_weight
initial_crop_size = frcnn_config.initial_crop_size
maxpool_kernel_size = frcnn_config.maxpool_kernel_size
maxpool_stride = frcnn_config.maxpool_stride
second_stage_box_predictor = box_predictor_builder.build(
hyperparams_builder.build,
frcnn_config.second_stage_box_predictor,
is_training=is_training,
num_classes=num_classes)
second_stage_batch_size = frcnn_config.second_stage_batch_size
second_stage_balance_fraction = frcnn_config.second_stage_balance_fraction
(second_stage_non_max_suppression_fn, second_stage_score_conversion_fn
) = post_processing_builder.build(frcnn_config.second_stage_post_processing)
second_stage_localization_loss_weight = (
frcnn_config.second_stage_localization_loss_weight)
second_stage_classification_loss = (
losses_builder.build_faster_rcnn_classification_loss(
frcnn_config.second_stage_classification_loss))
second_stage_classification_loss_weight = (
frcnn_config.second_stage_classification_loss_weight)
second_stage_mask_prediction_loss_weight = (
frcnn_config.second_stage_mask_prediction_loss_weight)
hard_example_miner = None
if frcnn_config.HasField('hard_example_miner'):
hard_example_miner = losses_builder.build_hard_example_miner(
frcnn_config.hard_example_miner,
second_stage_classification_loss_weight,
second_stage_localization_loss_weight)
common_kwargs = {
'is_training': is_training,
'num_classes': num_classes,
'image_resizer_fn': image_resizer_fn,
'feature_extractor': feature_extractor,
'number_of_stages': number_of_stages,
'first_stage_anchor_generator': first_stage_anchor_generator,
'first_stage_atrous_rate': first_stage_atrous_rate,
'first_stage_box_predictor_arg_scope':
first_stage_box_predictor_arg_scope,
'first_stage_box_predictor_kernel_size':
first_stage_box_predictor_kernel_size,
'first_stage_box_predictor_depth': first_stage_box_predictor_depth,
'first_stage_minibatch_size': first_stage_minibatch_size,
'first_stage_positive_balance_fraction':
first_stage_positive_balance_fraction,
'first_stage_nms_score_threshold': first_stage_nms_score_threshold,
'first_stage_nms_iou_threshold': first_stage_nms_iou_threshold,
'first_stage_max_proposals': first_stage_max_proposals,
'first_stage_localization_loss_weight': first_stage_loc_loss_weight,
'first_stage_objectness_loss_weight': first_stage_obj_loss_weight,
'second_stage_batch_size': second_stage_batch_size,
'second_stage_balance_fraction': second_stage_balance_fraction,
'second_stage_non_max_suppression_fn':
second_stage_non_max_suppression_fn,
'second_stage_score_conversion_fn': second_stage_score_conversion_fn,
'second_stage_localization_loss_weight':
second_stage_localization_loss_weight,
'second_stage_classification_loss':
second_stage_classification_loss,
'second_stage_classification_loss_weight':
second_stage_classification_loss_weight,
'hard_example_miner': hard_example_miner,
'add_summaries': add_summaries}
if isinstance(second_stage_box_predictor, box_predictor.RfcnBoxPredictor):
return rfcn_meta_arch.RFCNMetaArch(
second_stage_rfcn_box_predictor=second_stage_box_predictor,
**common_kwargs)
else:
return faster_rcnn_meta_arch.FasterRCNNMetaArch(
initial_crop_size=initial_crop_size,
maxpool_kernel_size=maxpool_kernel_size,
maxpool_stride=maxpool_stride,
second_stage_mask_rcnn_box_predictor=second_stage_box_predictor,
second_stage_mask_prediction_loss_weight=(
second_stage_mask_prediction_loss_weight),
**common_kwargs)
def _build_ssd_model(ssd_config, is_training, add_summaries):
"""Builds an SSD detection model based on the model config.
Args:
ssd_config: A ssd.proto object containing the config for the desired
SSDMetaArch.
is_training: True if this model is being built for training purposes.
add_summaries: Whether to add tf summaries in the model.
Returns:
SSDMetaArch based on the config.
Raises:
ValueError: If ssd_config.type is not recognized (i.e. not registered in
model_class_map).
"""
num_classes = ssd_config.num_classes
# Feature extractor
feature_extractor = _build_ssd_feature_extractor(
feature_extractor_config=ssd_config.feature_extractor,
freeze_batchnorm=ssd_config.freeze_batchnorm,
is_training=is_training)
box_coder = box_coder_builder.build(ssd_config.box_coder)
matcher = matcher_builder.build(ssd_config.matcher)
region_similarity_calculator = sim_calc.build(
ssd_config.similarity_calculator)
encode_background_as_zeros = ssd_config.encode_background_as_zeros
negative_class_weight = ssd_config.negative_class_weight
anchor_generator = anchor_generator_builder.build(
ssd_config.anchor_generator)
if feature_extractor.is_keras_model:
ssd_box_predictor = box_predictor_builder.build_keras(
conv_hyperparams_fn=hyperparams_builder.KerasLayerHyperparams,
freeze_batchnorm=ssd_config.freeze_batchnorm,
inplace_batchnorm_update=False,
num_predictions_per_location_list=anchor_generator
.num_anchors_per_location(),
box_predictor_config=ssd_config.box_predictor,
is_training=is_training,
num_classes=num_classes,
add_background_class=ssd_config.add_background_class)
else:
ssd_box_predictor = box_predictor_builder.build(
hyperparams_builder.build, ssd_config.box_predictor, is_training,
num_classes, ssd_config.add_background_class)
image_resizer_fn = image_resizer_builder.build(ssd_config.image_resizer)
non_max_suppression_fn, score_conversion_fn = post_processing_builder.build(
ssd_config.post_processing)
(classification_loss, localization_loss, classification_weight,
localization_weight, hard_example_miner,
random_example_sampler) = losses_builder.build(ssd_config.loss)
normalize_loss_by_num_matches = ssd_config.normalize_loss_by_num_matches
normalize_loc_loss_by_codesize = ssd_config.normalize_loc_loss_by_codesize
weight_regression_loss_by_score = (ssd_config.weight_regression_loss_by_score)
target_assigner_instance = target_assigner.TargetAssigner(
region_similarity_calculator,
matcher,
box_coder,
negative_class_weight=negative_class_weight,
weight_regression_loss_by_score=weight_regression_loss_by_score)
expected_classification_loss_under_sampling = None
if ssd_config.use_expected_classification_loss_under_sampling:
expected_classification_loss_under_sampling = functools.partial(
ops.expected_classification_loss_under_sampling,
min_num_negative_samples=ssd_config.min_num_negative_samples,
desired_negative_sampling_ratio=ssd_config.
desired_negative_sampling_ratio)
ssd_meta_arch_fn = ssd_meta_arch.SSDMetaArch
return ssd_meta_arch_fn(
is_training=is_training,
anchor_generator=anchor_generator,
box_predictor=ssd_box_predictor,
box_coder=box_coder,
feature_extractor=feature_extractor,
encode_background_as_zeros=encode_background_as_zeros,
image_resizer_fn=image_resizer_fn,
non_max_suppression_fn=non_max_suppression_fn,
score_conversion_fn=score_conversion_fn,
classification_loss=classification_loss,
localization_loss=localization_loss,
classification_loss_weight=classification_weight,
localization_loss_weight=localization_weight,
normalize_loss_by_num_matches=normalize_loss_by_num_matches,
hard_example_miner=hard_example_miner,
target_assigner_instance=target_assigner_instance,
add_summaries=add_summaries,
normalize_loc_loss_by_codesize=normalize_loc_loss_by_codesize,
freeze_batchnorm=ssd_config.freeze_batchnorm,
inplace_batchnorm_update=ssd_config.inplace_batchnorm_update,
add_background_class=ssd_config.add_background_class,
random_example_sampler=random_example_sampler,
expected_classification_loss_under_sampling=
expected_classification_loss_under_sampling)
def _build_sin_model(sin_config, is_training):
"""Builds a SIN detection model based on the model config.
Args:
sin_config: A faster_rcnn.proto object containing the config for the
desired FasterRCNNMetaArch or RFCNMetaArch.
is_training: True if this model is being built for training purposes.
Returns:
SINMetaArch based on the config.
Raises:
ValueError: If sin_config.type is not recognized (i.e. not registered in
model_class_map).
"""
num_classes = sin_config.num_classes
image_resizer_fn = image_resizer_builder.build(sin_config.image_resizer)
feature_extractor = _build_faster_rcnn_feature_extractor(
sin_config.feature_extractor, is_training, reuse_weights=tf.AUTO_REUSE)
fc_hyperparams = hyperparams_builder.build(
sin_config.second_stage_box_predictor.sin_box_predictor.fc_hyperparams,
is_training)
first_stage_only = sin_config.first_stage_only
first_stage_anchor_generator = anchor_generator_builder.build(
sin_config.first_stage_anchor_generator)
first_stage_atrous_rate = sin_config.first_stage_atrous_rate
first_stage_box_predictor_arg_scope = hyperparams_builder.build(
sin_config.first_stage_box_predictor_conv_hyperparams, is_training)
first_stage_box_predictor_kernel_size = (
sin_config.first_stage_box_predictor_kernel_size)
first_stage_box_predictor_depth = sin_config.first_stage_box_predictor_depth
first_stage_minibatch_size = sin_config.first_stage_minibatch_size
first_stage_positive_balance_fraction = (
sin_config.first_stage_positive_balance_fraction)
first_stage_nms_score_threshold = sin_config.first_stage_nms_score_threshold
first_stage_nms_iou_threshold = sin_config.first_stage_nms_iou_threshold
first_stage_max_proposals = sin_config.first_stage_max_proposals
first_stage_loc_loss_weight = (
sin_config.first_stage_localization_loss_weight)
first_stage_obj_loss_weight = sin_config.first_stage_objectness_loss_weight
initial_crop_size = sin_config.initial_crop_size
maxpool_kernel_size = sin_config.maxpool_kernel_size
maxpool_stride = sin_config.maxpool_stride
second_stage_box_predictor = box_predictor_builder.build(
hyperparams_builder.build,
sin_config.second_stage_box_predictor,
is_training=is_training,
num_classes=num_classes)
second_stage_batch_size = sin_config.second_stage_batch_size
second_stage_balance_fraction = sin_config.second_stage_balance_fraction
(second_stage_non_max_suppression_fn, second_stage_score_conversion_fn
) = post_processing_builder.build(sin_config.second_stage_post_processing)
second_stage_localization_loss_weight = (
sin_config.second_stage_localization_loss_weight)
second_stage_classification_loss = (
losses_builder.build_faster_rcnn_classification_loss(
sin_config.second_stage_classification_loss))
second_stage_classification_loss_weight = (
sin_config.second_stage_classification_loss_weight)
second_stage_mask_prediction_loss_weight = (
sin_config.second_stage_mask_prediction_loss_weight)
hard_example_miner = None
if sin_config.HasField('hard_example_miner'):
hard_example_miner = losses_builder.build_hard_example_miner(
sin_config.hard_example_miner,
second_stage_classification_loss_weight,
second_stage_localization_loss_weight)
common_kwargs = {
'is_training': is_training,
'num_classes': num_classes,
'image_resizer_fn': image_resizer_fn,
'feature_extractor': feature_extractor,
'fc_hyperparams': fc_hyperparams,
'first_stage_only': first_stage_only,
'first_stage_anchor_generator': first_stage_anchor_generator,
'first_stage_atrous_rate': first_stage_atrous_rate,
'first_stage_box_predictor_arg_scope':
first_stage_box_predictor_arg_scope,
'first_stage_box_predictor_kernel_size':
first_stage_box_predictor_kernel_size,
'first_stage_box_predictor_depth': first_stage_box_predictor_depth,
'first_stage_minibatch_size': first_stage_minibatch_size,
'first_stage_positive_balance_fraction':
first_stage_positive_balance_fraction,
'first_stage_nms_score_threshold': first_stage_nms_score_threshold,
'first_stage_nms_iou_threshold': first_stage_nms_iou_threshold,
'first_stage_max_proposals': first_stage_max_proposals,
'first_stage_localization_loss_weight': first_stage_loc_loss_weight,
'first_stage_objectness_loss_weight': first_stage_obj_loss_weight,
'second_stage_batch_size': second_stage_batch_size,
'second_stage_balance_fraction': second_stage_balance_fraction,
'second_stage_non_max_suppression_fn':
second_stage_non_max_suppression_fn,
'second_stage_score_conversion_fn': second_stage_score_conversion_fn,
'second_stage_localization_loss_weight':
second_stage_localization_loss_weight,
'second_stage_classification_loss':
second_stage_classification_loss,
'second_stage_classification_loss_weight':
second_stage_classification_loss_weight,
'hard_example_miner': hard_example_miner}
if isinstance(second_stage_box_predictor, box_predictor.RfcnBoxPredictor):
return rfcn_meta_arch.RFCNMetaArch(
second_stage_rfcn_box_predictor=second_stage_box_predictor,
**common_kwargs)
else:
return sin_meta_arch.SINMetaArch(
initial_crop_size=initial_crop_size,
maxpool_kernel_size=maxpool_kernel_size,
maxpool_stride=maxpool_stride,
second_stage_box_predictor=second_stage_box_predictor,
second_stage_mask_prediction_loss_weight=(
second_stage_mask_prediction_loss_weight),
**common_kwargs)
def _build_ssd_model(ssd_config, is_training):
"""Builds an SSD detection model based on the model config.
Args:
ssd_config: A ssd.proto object containing the config for the desired
SSDMetaArch.
is_training: True if this model is being built for training purposes.
Returns:
SSDMetaArch based on the config.
Raises:
ValueError: If ssd_config.type is not recognized (i.e. not registered in
model_class_map).
"""
num_classes = ssd_config.num_classes #number of clases
# Feature extractor
feature_extractor = _build_ssd_feature_extractor(ssd_config.feature_extractor, #we use ssd_mobilenet_v1 as the feature extractor
is_training) #set the class in ssd_mobilenr_v1_feature_extractor amd ssd_meta+arch.py
#when taking the regression loss we are working with some transorfmation. That means our predictors will predict 4 cordinates and those codinates should be regressed with some kind embedding which was made with ground truth boxes and default boxes , then after getting those we docode them for real images
box_coder = box_coder_builder.build(ssd_config.box_coder) #set en encoding w.r.t ground truth boxes and achor boxes . The output creating with this object will then regressed with the predicted onece. chenck equation 2 in the ssd paper
matcher = matcher_builder.build(ssd_config.matcher) #matching the predicted to ground trunth- Builds a matcher object based on the matcher config
#in obove object matching is done with default boxes and ground truth boxes , that's how xij value in the paper obtained .
region_similarity_calculator = sim_calc.build( #how to calculate the similarity parameter is iou .
ssd_config.similarity_calculator)
ssd_box_predictor = box_predictor_builder.build(hyperparams_builder.build, #This will take care of the convolutional kernal
ssd_config.box_predictor,
is_training, num_classes) #this returns a box_predictor object
anchor_generator = anchor_generator_builder.build( #pass an instance or object where we can create ancho boxes for differen featuremaps
ssd_config.anchor_generator)
image_resizer_fn = image_resizer_builder.build(ssd_config.image_resizer) #this is imortatnt we use fixed_shape_resizer
non_max_suppression_fn, score_conversion_fn = post_processing_builder.build( #this is to work with NMS supression output
ssd_config.post_processing) #score conversion function will convert logits to probabilities
(classification_loss, localization_loss, classification_weight,
localization_weight,
hard_example_miner) = losses_builder.build(ssd_config.loss) #now the loss for hard examples these outputs are objects
normalize_loss_by_num_matches = ssd_config.normalize_loss_by_num_matches # we devide by the matching acnhorboxes
return ssd_meta_arch.SSDMetaArch( #here we initialized a object of ssd_meta_arch which will be used in trainign
is_training,
anchor_generator,
ssd_box_predictor,
box_coder,
feature_extractor,
matcher,
region_similarity_calculator,
image_resizer_fn,
non_max_suppression_fn,
score_conversion_fn,
classification_loss,
localization_loss,
classification_weight,
localization_weight,
normalize_loss_by_num_matches,
hard_example_miner)
