Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

[Relay][OP] Support NMSv4 #6085

Merged
merged 2 commits into from
Jul 27, 2020
Merged

[Relay][OP] Support NMSv4 #6085

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
3554461
[Relay][OP] Support NMSv4 ingestion from TF.
csullivan Jul 17, 2020
77b5c3d
Merge branch 'master' of github.com:apache/incubator-tvm into cs/nmsv4
csullivan Jul 17, 2020
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 5 additions & 1 deletion python/tvm/relay/frontend/tensorflow.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -637,10 +637,11 @@ def _impl(inputs, attr, params, mod):
iou_threshold = np.atleast_1d(inputs[3].data.asnumpy())[0]
# score_threshold was introduced from V3
score_threshold = np.atleast_1d(inputs[4].data.asnumpy())[0] if len(inputs) > 4 else 0.0
pad_output = 'pad_to_max_output_size'

# Generate data with shape (1, num_anchors, 5)
scores = AttrCvt(op_name="expand_dims",
ignores=['T_threshold'],
ignores=['T_threshold', pad_output],
Copy link
Contributor

@kevinthesun kevinthesun Jul 18, 2020 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We want to do padding if number of output boxes is smaller than max_output_size?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

NMSv4 semantics allow for the user to define whether or not to pad the output using pad_to_max_output_size=True/False, see the example I provided in the test. Some pre-trained TF models use NMSv4 with pad_to_max_output_size=True, in which case, yes, they expect padding of the indices if the number of boxes is less than max_output_size, as well as an additional scalar output specifying the number of valid boxes.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We need to handle True case instead of just ignoring it?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It's not ignored, this PR adds explicit support to __nms() for the pad_to_max_output_size=True case, see lines 671/2.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Got it. LGTM.

extras={'axis': -1, 'num_newaxis': 1})([inputs[1]], attr)
data = get_relay_op('concatenate')([scores, inputs[0]], -1)
data = get_relay_op('expand_dims')(data, 0, 1)
Expand All @@ -667,6 +668,8 @@ def _impl(inputs, attr, params, mod):
return_indices=True,
invalid_to_bottom=False)

if pad_output in attr and attr[pad_output]:
return nms_ret
# squeeze it, TF NMS is not batched
size = get_relay_op("squeeze")(nms_ret[1], axis=[1])
data_slice = get_relay_op("squeeze")(nms_ret[0], axis=[0])
Expand Down Expand Up @@ -2152,6 +2155,7 @@ def _impl(inputs, attr, params, mod):
'Neg' : AttrCvt('negative'),
'NonMaxSuppressionV2' : _nms(),
'NonMaxSuppressionV3' : _nms(),
'NonMaxSuppressionV4' : _nms(),
'NoOp' : _no_op(),
'NotEqual' : _broadcast('not_equal'),
'OneHot' : _one_hot(),
Expand Down
33 changes: 26 additions & 7 deletions tests/python/frontend/tensorflow/test_forward.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2032,12 +2032,31 @@ def _test_forward_nms_v3(bx_shape, score_shape, iou_threshold, score_threshold,
compare_tf_with_tvm([boxes, scores, max_output_size], ['in_data_1:0', 'in_data_2:0', 'in_data_3:0'],
'nms/NonMaxSuppressionV3:0', mode='debug')

def test_forward_nms_v3():
""" NonMaxSuppressionV3 """
_test_forward_nms_v3((5, 4), (5,), 0.7, 0.5, 5)
_test_forward_nms_v3((20, 4), (20,), 0.5, 0.6, 10)
_test_forward_nms_v3((1000, 4), (1000,), 0.3, 0.7, 1000)
_test_forward_nms_v3((2000, 4), (2000,), 0.4, 0.6, 7)
def _test_forward_nms_v4(bx_shape, score_shape, iou_threshold, score_threshold, out_size, dtype="float32"):
boxes = np.random.uniform(0, 10, size=bx_shape).astype(dtype)
scores = np.random.uniform(size=score_shape).astype(dtype)
max_output_size = np.int32(out_size)
tf.reset_default_graph()
in_data_1 = tf.placeholder(dtype, boxes.shape, name="in_data_1")
in_data_2 = tf.placeholder(dtype, scores.shape, name="in_data_2")
in_data_3 = tf.placeholder(tf.int32, name="in_data_3")
indices_padded, num_valid = tf.image.non_max_suppression_padded(boxes=in_data_1, scores=in_data_2, max_output_size=in_data_3,
iou_threshold=iou_threshold, score_threshold=score_threshold, name="nms", pad_to_max_output_size=True)
num_valid = tf.reshape(num_valid,shape=(-1,))
indices_padded = tf.reshape(indices_padded, shape=(-1,))
tf.slice(indices_padded, tf.constant([0]), num_valid, name="SlicedIndices")
compare_tf_with_tvm([boxes, scores, max_output_size], ['in_data_1:0', 'in_data_2:0', 'in_data_3:0'],
['nms/NonMaxSuppressionV4:1', "SlicedIndices:0"], mode='vm')
compare_tf_with_tvm([boxes, scores, max_output_size], ['in_data_1:0', 'in_data_2:0', 'in_data_3:0'],
['nms/NonMaxSuppressionV4:1', "SlicedIndices:0"], mode='debug')

def test_forward_nms():
""" NonMaxSuppressionV3,4 """
for _test_forward_nms in [_test_forward_nms_v3, _test_forward_nms_v4]:
_test_forward_nms((5, 4), (5,), 0.7, 0.5, 5)
_test_forward_nms((20, 4), (20,), 0.5, 0.6, 10)
_test_forward_nms((1000, 4), (1000,), 0.3, 0.7, 1000)
_test_forward_nms((2000, 4), (2000,), 0.4, 0.6, 7)


#######################################################################
Expand Down Expand Up @@ -3868,7 +3887,7 @@ def lstm_cell():
test_forward_truncatemod()
test_forward_one_hot()
test_forward_atan2()
test_forward_nms_v3()
test_forward_nms()

# Activations
test_forward_sigmoid()
Expand Down