def build(box_coder_config):
if not isinstance(box_coder_config, box_coder_pb2.BoxCoder):
raise ValueError(
'box_coder_config not of type box_coder_pb2.BoxCoder.')
if box_coder_config.WhichOneof(
'box_coder_oneof') == 'faster_rcnn_box_coder':
return faster_rcnn_box_coder.FasterRcnnBoxCoder(scale_factors=[
box_coder_config.faster_rcnn_box_coder.y_scale,
box_coder_config.faster_rcnn_box_coder.x_scale,
box_coder_config.faster_rcnn_box_coder.height_scale,
box_coder_config.faster_rcnn_box_coder.width_scale
])
if box_coder_config.WhichOneof('box_coder_oneof') == 'keypoint_box_coder':
return keypoint_box_coder.KeypointBoxCoder(
box_coder_config.keypoint_box_coder.num_keypoints,
scale_factors=[
box_coder_config.keypoint_box_coder.y_scale,
box_coder_config.keypoint_box_coder.x_scale,
box_coder_config.keypoint_box_coder.height_scale,
box_coder_config.keypoint_box_coder.width_scale
])
if (box_coder_config.WhichOneof('box_coder_oneof') ==
'mean_stddev_box_coder'):
return mean_stddev_box_coder.MeanStddevBoxCoder(
stddev=box_coder_config.mean_stddev_box_coder.stddev)
if box_coder_config.WhichOneof('box_coder_oneof') == 'square_box_coder':
return square_box_coder.SquareBoxCoder(scale_factors=[
box_coder_config.square_box_coder.y_scale,
box_coder_config.square_box_coder.x_scale,
box_coder_config.square_box_coder.length_scale
])
raise ValueError('Empty box coder.')
def graph_fn(boxes, keypoints, anchors):
boxes = box_list.BoxList(boxes)
boxes.add_field(fields.BoxListFields.keypoints, keypoints)
anchors = box_list.BoxList(anchors)
coder = keypoint_box_coder.KeypointBoxCoder(num_keypoints)
rel_codes = coder.encode(boxes, anchors)
return rel_codes
def graph_fn(rel_codes, anchors):
scale_factors = [2, 3, 4, 5]
anchors = box_list.BoxList(anchors)
coder = keypoint_box_coder.KeypointBoxCoder(
num_keypoints, scale_factors=scale_factors)
boxes = coder.decode(rel_codes, anchors)
return boxes.get(), boxes.get_field(fields.BoxListFields.keypoints)
def test_very_small_width_nan_after_encoding(self):
boxes = [[10., 10., 10.0000001, 20.]]
keypoints = [[[10., 10.], [10.0000001, 20.]]]
anchors = [[15., 12., 30., 18.]]
expected_rel_codes = [[-0.833333, 0., -21.128731, 0.510826,
-0.833333, -0.833333, -0.833333, 0.833333]]
boxes = box_list.BoxList(tf.constant(boxes))
boxes.add_field(fields.BoxListFields.keypoints, tf.constant(keypoints))
anchors = box_list.BoxList(tf.constant(anchors))
coder = keypoint_box_coder.KeypointBoxCoder(2)
rel_codes = coder.encode(boxes, anchors)
with self.test_session() as sess:
rel_codes_out, = sess.run([rel_codes])
self.assertAllClose(rel_codes_out, expected_rel_codes)
def build(box_coder_config):
"""Builds a box coder object based on the box coder config.
Args:
box_coder_config: A box_coder.proto object containing the config for the
desired box coder.
Returns:
BoxCoder based on the config.
Raises:
ValueError: On empty box coder proto.
"""
print("CONFIG:{}".format(box_coder_config))
print("PB2:{}".format(box_coder_pb2.BoxCoder))
if not isinstance(box_coder_config, box_coder_pb2.BoxCoder):
print("CONFIG:{}".format(box_coder_config))
print("PB2:{}".format(box_coder_pb2.BoxCoder))
raise ValueError(
'box_coder_config not of type box_coder_pb2.BoxCoder.')
if box_coder_config.WhichOneof(
'box_coder_oneof') == 'faster_rcnn_box_coder':
return faster_rcnn_box_coder.FasterRcnnBoxCoder(scale_factors=[
box_coder_config.faster_rcnn_box_coder.y_scale,
box_coder_config.faster_rcnn_box_coder.x_scale,
box_coder_config.faster_rcnn_box_coder.height_scale,
box_coder_config.faster_rcnn_box_coder.width_scale
])
if box_coder_config.WhichOneof('box_coder_oneof') == 'keypoint_box_coder':
return keypoint_box_coder.KeypointBoxCoder(
box_coder_config.keypoint_box_coder.num_keypoints,
scale_factors=[
box_coder_config.keypoint_box_coder.y_scale,
box_coder_config.keypoint_box_coder.x_scale,
box_coder_config.keypoint_box_coder.height_scale,
box_coder_config.keypoint_box_coder.width_scale
])
if (box_coder_config.WhichOneof('box_coder_oneof') ==
'mean_stddev_box_coder'):
return mean_stddev_box_coder.MeanStddevBoxCoder()
if box_coder_config.WhichOneof('box_coder_oneof') == 'square_box_coder':
return square_box_coder.SquareBoxCoder(scale_factors=[
box_coder_config.square_box_coder.y_scale,
box_coder_config.square_box_coder.x_scale,
box_coder_config.square_box_coder.length_scale
])
raise ValueError('Empty box coder.')
def graph_fn(anchor_means, groundtruth_box_corners,
groundtruth_keypoints):
similarity_calc = region_similarity_calculator.IouSimilarity()
matcher = argmax_matcher.ArgMaxMatcher(matched_threshold=0.5,
unmatched_threshold=0.5)
box_coder = keypoint_box_coder.KeypointBoxCoder(
num_keypoints=6, scale_factors=[10.0, 10.0, 5.0, 5.0])
target_assigner = targetassigner.TargetAssigner(
similarity_calc, matcher, box_coder, unmatched_cls_target=None)
anchors_boxlist = box_list.BoxList(anchor_means)
groundtruth_boxlist = box_list.BoxList(groundtruth_box_corners)
groundtruth_boxlist.add_field(fields.BoxListFields.keypoints,
groundtruth_keypoints)
result = target_assigner.assign(anchors_boxlist, groundtruth_boxlist)
(cls_targets, cls_weights, reg_targets, reg_weights, _) = result
return (cls_targets, cls_weights, reg_targets, reg_weights)
def test_get_correct_relative_codes_after_encoding(self):
boxes = [[10., 10., 20., 15.], [0.2, 0.1, 0.5, 0.4]]
keypoints = [[[15., 12.], [10., 15.]], [[0.5, 0.3], [0.2, 0.4]]]
num_keypoints = len(keypoints[0])
anchors = [[15., 12., 30., 18.], [0.1, 0.0, 0.7, 0.9]]
expected_rel_codes = [[
-0.5, -0.416666, -0.405465, -0.182321, -0.5, -0.5, -0.833333, 0.
],
[
-0.083333, -0.222222, -0.693147, -1.098612,
0.166667, -0.166667, -0.333333, -0.055556
]]
boxes = box_list.BoxList(tf.constant(boxes))
boxes.add_field(fields.BoxListFields.keypoints, tf.constant(keypoints))
anchors = box_list.BoxList(tf.constant(anchors))
coder = keypoint_box_coder.KeypointBoxCoder(num_keypoints)
rel_codes = coder.encode(boxes, anchors)
with self.test_session() as sess:
rel_codes_out, = sess.run([rel_codes])
self.assertAllClose(rel_codes_out, expected_rel_codes)
def test_get_correct_relative_codes_after_encoding_with_scaling(self):
boxes = [[10., 10., 20., 15.], [0.2, 0.1, 0.5, 0.4]]
keypoints = [[[15., 12.], [10., 15.]], [[0.5, 0.3], [0.2, 0.4]]]
num_keypoints = len(keypoints[0])
anchors = [[15., 12., 30., 18.], [0.1, 0.0, 0.7, 0.9]]
scale_factors = [2, 3, 4, 5]
expected_rel_codes = [[
-1., -1.25, -1.62186, -0.911608, -1.0, -1.5, -1.666667, 0.
],
[
-0.166667, -0.666667, -2.772588, -5.493062,
0.333333, -0.5, -0.666667, -0.166667
]]
boxes = box_list.BoxList(tf.constant(boxes))
boxes.add_field(fields.BoxListFields.keypoints, tf.constant(keypoints))
anchors = box_list.BoxList(tf.constant(anchors))
coder = keypoint_box_coder.KeypointBoxCoder(
num_keypoints, scale_factors=scale_factors)
rel_codes = coder.encode(boxes, anchors)
with self.test_session() as sess:
rel_codes_out, = sess.run([rel_codes])
self.assertAllClose(rel_codes_out, expected_rel_codes)
def create_target_assigner(reference,
stage=None,
negative_class_weight=1.0,
use_matmul_gather=False):
"""Factory function for creating standard target assigners.
Args:
reference: string referencing the type of TargetAssigner.
stage: string denoting stage: {proposal, detection}.
negative_class_weight: classification weight to be associated to negative
anchors (default: 1.0)
use_matmul_gather: whether to use matrix multiplication based gather which
are better suited for TPUs.
Returns:
TargetAssigner: desired target assigner.
Raises:
ValueError: if combination reference+stage is invalid.
"""
if reference == 'Multibox' and stage == 'proposal':
similarity_calc = sim_calc.NegSqDistSimilarity()
matcher = bipartite_matcher.GreedyBipartiteMatcher()
box_coder = mean_stddev_box_coder.MeanStddevBoxCoder()
elif reference == 'FasterRCNN' and stage == 'proposal':
similarity_calc = sim_calc.IouSimilarity()
matcher = argmax_matcher.ArgMaxMatcher(
matched_threshold=0.7,
unmatched_threshold=0.3,
force_match_for_each_row=True,
use_matmul_gather=use_matmul_gather)
box_coder = faster_rcnn_box_coder.FasterRcnnBoxCoder(
scale_factors=[10.0, 10.0, 5.0, 5.0])
elif reference == 'FasterRCNN' and stage == 'detection':
similarity_calc = sim_calc.IouSimilarity()
# Uses all proposals with IOU < 0.5 as candidate negatives.
matcher = argmax_matcher.ArgMaxMatcher(
matched_threshold=0.5,
negatives_lower_than_unmatched=True,
use_matmul_gather=use_matmul_gather)
box_coder = faster_rcnn_box_coder.FasterRcnnBoxCoder(
scale_factors=[10.0, 10.0, 5.0, 5.0])
elif reference == 'FasterRCNN' and stage == 'keypoints':
similarity_calc = sim_calc.IouSimilarity()
# Uses all proposals with IOU < 0.5 as candidate negatives.
matcher = argmax_matcher.ArgMaxMatcher(
matched_threshold=0.5,
negatives_lower_than_unmatched=True,
use_matmul_gather=use_matmul_gather)
box_coder = keypoint_box_coder.KeypointBoxCoder(
6, scale_factors=[10.0, 10.0, 5.0, 5.0])
elif reference == 'FastRCNN':
similarity_calc = sim_calc.IouSimilarity()
matcher = argmax_matcher.ArgMaxMatcher(
matched_threshold=0.5,
unmatched_threshold=0.1,
force_match_for_each_row=False,
negatives_lower_than_unmatched=False,
use_matmul_gather=use_matmul_gather)
box_coder = faster_rcnn_box_coder.FasterRcnnBoxCoder()
else:
raise ValueError('No valid combination of reference and stage.')
return TargetAssigner(similarity_calc,
matcher,
box_coder,
negative_class_weight=negative_class_weight)
