Skip to content

M
Models

曾经的那一瞬间 / Models
大约 1 年 前同步成功

通知 1
Star 0
Fork 0
M
Models

体验新版 GitCode，发现更多精彩内容 >>

提交 5938a718 编写于 作者: A A. Unique TensorFlower
浏览文件
操作

Internal change 

PiperOrigin-RevId: 491555987
上级 c5662b16
隐藏空白更改
内联 并排
Showing with 519 addition and 480 deletion
official/projects/edgetpu/vision/modeling/custom_layers.py
浏览文件 @ 5938a718
......@@ -14,10 +14,9 @@
"""Customized keras layers used in the EdgeTPU models."""
from collections.abc import Iterable, MutableMapping, Sequence
from collections.abc import MutableMapping
import inspect
from typing import Any, Optional, Union
import tensorflow as tf
from official.modeling import tf_utils
......@@ -479,310 +478,3 @@ class ArgmaxKerasLayer(tf.keras.layers.Layer):
axis=self.axis,
output_type=self.output_type,
name=self.name)
_or = tf.maximum
_and = tf.minimum
_reduce_or = tf.reduce_max
def _tensor_sum_vectors(a, b):
a = tf.tile(tf.reshape(a, [1, -1, 1, a.shape[-1]]), [1, 1, a.shape[-1], 1])
b = tf.tile(tf.reshape(b, [1, -1, a.shape[-1], 1]), [1, 1, 1, a.shape[-1]])
return a + b
def _tensor_product_iou(boxes):
"""Computes pairwise IOU.
Reason to use 4-D tensors is to follow TPU compiler preference.
Args:
boxes: A 2-D float `Tensor` of shape `[num_boxes, 4]`.
Returns:
A 4-D float `Tensor` of shape `[1, 1, num_boxes, num_boxes]` containing
pairwise IOU.
"""
boxes_size = boxes.shape[-2]
# Code below will do frequent operands broadcasting.
# TPU compiler has (empirically) less issues broadcasting if
# - batch (first) dimension is 1. (Special consideration sharding)
# - there are 4 dimensions. (Standard traversal mapping)
# - last dimension is not 1. (Structure alignment)
tpu_friendly_shape = [1, -1, 1, boxes_size]
bottom, left, top, right = (
tf.reshape(side, tpu_friendly_shape) for side in tf.split(boxes, 4, -1))
height, width = top - bottom, right - left
area = height * width
area_sum = _tensor_sum_vectors(area, area)
bottom_pad, left_pad, top_pad, right_pad = (
tf.nn.relu(_tensor_sum_vectors(x, -x))
for x in (-bottom, -left, top, right))
height_pad, width_pad = bottom_pad + top_pad, left_pad + right_pad
intersection = tf.nn.relu(height - height_pad) * tf.nn.relu(width - width_pad)
union = area_sum - intersection
iou = tf.math.divide(intersection, union + _same(union))
return iou
def _greater(x):
"""Avoid non lowerable layers in boolean comparison.
Logical operation results in tensor of boolean type. However in serving such
a tensors cannot be cast to values because of NNAPI specs.
`tf.where` operation result in `select` instruction lowering, which not runs
well on all generations of edge-tpus.
Args:
x: any numeric tensor.
Returns:
tf.where(x > tf.zero_like(x), tf.one_like(x), tf.zero_like(x))
"""
x_clip = tf.minimum(tf.nn.relu(x), tf.constant(1, dtype=x.dtype))
return -tf.math.floor(-x_clip)
def _same(x):
"""Avoid non lowerable layers in boolean equality.
Logical operation results in tensor of boolean type. However in serving such
a tensors cannot be cast to values because of NNAPI specs.
`tf.where` operation result in `select` instruction lowering, which not runs
well on all generations of edge-tpus.
Args:
x: any numeric tensor.
Returns:
tf.where(x == tf.zero_like(x), tf.one_like(x), tf.zero_like(x))
"""
x_clip = tf.minimum(tf.abs(x), tf.constant(1, dtype=x.dtype))
return tf.constant(1, dtype=x.dtype) + tf.math.floor(-x_clip)
def shard_tensors(axis: int, block_size: int,
*tensors: tf.Tensor) -> Iterable[Sequence[tf.Tensor]]:
"""Consistently splits multiple tensors sharding-style.
Args:
axis: axis to be used to split tensors
block_size: block size to split tensors.
*tensors: list of tensors.
Returns:
List of shards, each shard has exactly one peace of each input tesnor.
Raises:
ValueError: if input tensors has different size of sharded dimension.
"""
for validate_axis in range(axis + 1):
consistent_length: int = tensors[0].shape[validate_axis]
for tensor in tensors:
if tensor.shape[validate_axis] != consistent_length:
raise ValueError('Inconsistent shapes in shard_tensors: first is '
f'{tensors[0].shape} and other is {tensor.shape}')
batch_size: int = tensors[0].shape[axis]
if block_size >= batch_size:
return [tensors]
else:
blocks = batch_size // block_size
remainder = batch_size % block_size
if remainder:
tensor_parts = []
for tensor in tensors:
shape: tf.TensorShape = tensor.shape
body: tf.Tensor = tf.slice(tensor, [0] * len(shape), [
size if i != axis else blocks * block_size
for i, size in enumerate(shape)
])
tail: tf.Tensor = tf.slice(tensor, [
0 if i != axis else (blocks * block_size)
for i, _ in enumerate(shape)
], [
size if i != axis else (size - blocks * block_size)
for i, size in enumerate(shape)
])
tensor_parts.append(tf.split(body, blocks, axis) + [tail])
return zip(*tensor_parts)
else:
return zip(*[tf.split(tensor, blocks, axis) for tensor in tensors])
# TODO(b/258007436): Number is based on existing compiler limitations while
# running bf16 NMS on edgetpu. Remove manual sharing when compiler issue will be
# fixed.
_RECOMMENDED_NMS_MEMORY = 360000
def non_max_suppression_padded(boxes: tf.Tensor,
scores: tf.Tensor,
output_size: int,
iou_threshold: float = 0.5) -> tf.Tensor:
"""Selects a subset of boxes which have highest score among IOU-similar boxes.
Prunes away boxes that have high intersection-over-union (IOU) overlap
with boxes having higher score. Boxes are supplied as `[y1, x1, y2, x2]`,
where `(y1, x1)` and `(y2, x2)` are the coordinates of any diagonal pair of
box corners. Note that this algorithm is agnostic to the coordinate system.
Thus translating or reflections of the coordinate system result in the same
boxes being selected by the algorithm. The output of this operation is a
set of integers indexing into the input collection of bounding boxes
representing the selected boxes.
Set will be returned padded on the right with `-1` values. The bounding
box coordinates corresponding to the selected indices can then be obtained
using the `tf.gather` operation. For example:
```python
selected_indices = vision.modeling.layers.non_max_suppression_padded(
boxes, scores, max_output_size, iou_threshold)
selected_boxes = tf.gather(boxes, selected_indices)
```
See following documetation for implementation details.
third_party/tensorflow_models/official/projects/edgetpu/vision/modeling/g3doc/non_max_suppression.md
Args:
boxes: A 2-D+ float `Tensor` of shape `[...batch_dims, num_boxes, 4]`.
scores: A 1-D+ float `Tensor` of shape `[...batch_dims, num_boxes]`
representing a single score corresponding to each box (each row of boxes).
output_size: A scalar integer `Tensor` representing the maximum number of
boxes to be selected by non-max suppression.
iou_threshold: A 0-D float tensor representing the threshold for deciding
whether boxes overlap too much with respect to IOU.
Returns:
A 1-D+ integer `Tensor` of shape `[...batch_dims, output_size]` representing
the selected indices from the boxes tensor and `-1` values for the padding.
"""
# Does partitioning job to help compiler converge with memory.
batch_shape = boxes.shape[:-2]
batch_size = tf.reduce_prod(batch_shape).numpy()
boxes_size, struct_size = boxes.shape[-2:]
boxes = tf.reshape(boxes, [batch_size, boxes_size, struct_size])
scores = tf.reshape(scores, [batch_size, boxes_size])
block = max(1, _RECOMMENDED_NMS_MEMORY // (boxes_size * boxes_size))
indices = []
for boxes_i, scores_i in shard_tensors(0, block, boxes, scores):
indices.append(
_non_max_suppression_as_is(boxes_i, scores_i, output_size,
iou_threshold))
indices = tf.concat(indices, axis=0)
return tf.reshape(indices, batch_shape + [output_size])
def _non_max_suppression_as_is(boxes: tf.Tensor,
scores: tf.Tensor,
output_size: int,
iou_threshold: float = 0.5) -> tf.Tensor:
"""Selects a subset of boxes which have highest score among IOU-similar boxes.
Args:
boxes: A 2-D+ float `Tensor` of shape `[...batch_dims, num_boxes, 4]`.
scores: A 1-D+ float `Tensor` of shape `[...batch_dims, num_boxes]`
representing a single score corresponding to each box (each row of boxes).
output_size: A scalar integer `Tensor` representing the maximum number of
boxes to be selected by non-max suppression.
iou_threshold: A 0-D float tensor representing the threshold for deciding
whether boxes overlap too much with respect to IOU.
Returns:
A 1-D+ integer `Tensor` of shape `[...batch_dims, output_size]` representing
the selected indices from the boxes tensor and `-1` values for the padding.
"""
batch_shape = boxes.shape[:-2]
batch_size = tf.reduce_prod(batch_shape).numpy()
boxes_size = boxes.shape[-2]
if boxes.shape[-1] != 4:
raise ValueError(f'Boxes shape ({boxes.shape}) last dimension must be 4 '
'to represent [y1, x1, y2, x2] boxes coordinates')
if scores.shape != boxes.shape[:-1]:
raise ValueError(f'Boxes shape ({boxes.shape}) and scores shape '
f'({scores.shape}) do not match.')
order = tf.range(boxes_size, dtype=tf.float32)
relative_order = _tensor_sum_vectors(order, -order)
relative_scores = _tensor_sum_vectors(scores, -scores)
similar = _greater(_tensor_product_iou(boxes) - iou_threshold)
worse = _greater(relative_scores)
same_later = _and(_same(relative_scores), _greater(relative_order))
similar_worse_or_same_later = _and(similar, _or(worse, same_later))
prunable = _reduce_or(similar_worse_or_same_later, axis=-1)
remaining = tf.constant(1.) - prunable
scores = tf.reshape(tf.exp(scores), [1, 1, batch_size, boxes_size])
remaining = tf.reshape(remaining, [1, 1, batch_size, boxes_size])
# top_k runs on TPU cores, let it happen, TPU tiles implementation is slower.
top_k = tf.math.top_k(scores * remaining, output_size)
indices = (
tf.cast(top_k.indices, top_k.values.dtype) * _greater(top_k.values) -
_same(top_k.values))
return tf.reshape(indices, batch_shape + [output_size])
def concat_and_top_k(
top_k: int, scores_pair: tuple[Optional[tf.Tensor], tf.Tensor],
*other_pairs: tuple[Optional[tf.Tensor], tf.Tensor]
) -> tuple[tf.Tensor, ...]:
"""Combines shards of top_k operation, when sharded along filtered dimension.
General idea is that sometimes top_k dimension is very large, while top_k is
moderately low. (Keep in mind sample of 15K pre-top_k dimension and 150 top_k)
In that case it is possible to break top_k input into groups significantly
larger than top_k and significatly lower than pre-top_l (Keep in mind 1500).
We do top_k over first 1500 elements, than join 150 remaining with new 1500
elements (1750 in total), repeat top_k. This function provides repeatedly used
method which will concat and top_k in that case.
For example with top_k = 2 and scores_pair = ([10, 6], [9, 8, 7]), output
scores will be [10, 9].
Other pairs are filtered using indexes generated from scores. This is a preaty
common case of filtering structure by its score.
For example with one extra pair of box per score:
top_k = 2
scores_pair = ([10, 6],
[9, 8, 7])
other_pairs = [([[0, 0, 10, 10], [0, 0, 6, 6]],
[[1, 1, 9, 9], [1, 1, 8, 8], [1, 1, 7, 7]])]
Output is:
([10, 9], [[0, 0, 10, 10], [1, 1, 9, 9]])
See also 'test_top_k_sharded_fusion' unit test with end to end example.
Args:
top_k: is top_k argument of sharded tf.math.top_k.
scores_pair: Tuple (<previous shards combination>, <additional shard>)
scores to be aggregated using top_k.
*other_pairs: Tuples (<previous shards combination>, <additional shard>)
other values to be aggregated using indexes of top_k scores.
Returns:
Tuple of scores based top_k aggregations with additional shards.
"""
scores, scores_shard = scores_pair
if other_pairs:
others, others_shard = zip(*other_pairs)
else:
others = others_shard = []
# Same as tf.rank, but avoiding tensor form for graph mode execution.
top_k_dim: int = len(scores_shard.shape) - 1
if scores is None:
# First shard becomes aggregation
scores = scores_shard
others = others_shard
else:
# Merge shard into agregation
scores = tf.concat([scores, scores_shard], top_k_dim)
others = [
tf.concat([other, other_shard], top_k_dim)
for other, other_shard in zip(others, others_shard)
]
# When shards are uneven some will be smaller than requested top_k
if scores.shape[top_k_dim] > top_k:
scores, indices = tf.nn.top_k(scores, top_k)
others = [
tf.gather(other, indices, axis=top_k_dim, batch_dims=top_k_dim)
for other in others
]
return scores, *others
official/projects/edgetpu/vision/modeling/custom_layers_test.py
浏览文件 @ 5938a718
......@@ -15,10 +15,8 @@
"""Tests for custom_layers."""
import itertools
from typing import Optional
from absl.testing import parameterized
import numpy as np
import tensorflow as tf
from official.projects.edgetpu.vision.modeling import custom_layers
......@@ -186,170 +184,3 @@ class ArgmaxTest(parameterized.TestCase, tf.test.TestCase):
test_output = custom_layers.argmax(
random_inputs, axis=axis, output_type=output_type)
self.assertAllEqual(control_output, test_output)
def random_boxes(shape):
a = tf.random.uniform(shape=shape+[2])
b = tf.random.uniform(shape=shape+[2])
l = tf.minimum(a, b)
u = tf.maximum(a, b)
return tf.concat([l, u], axis=-1)
def _maximum_activation_size(model):
max_size = 0
for layer in model.layers:
outputs = layer.output
if not isinstance(outputs, list):
outputs = [outputs]
for output in outputs:
if hasattr(output, 'shape'):
size = np.prod(output.shape)
max_size = max(max_size, size)
print('Layer', size, output.shape, layer.name)
return max_size
class NonMaxSuppressionTest(parameterized.TestCase, tf.test.TestCase):
def setUp(self):
super().setUp()
tf.random.set_seed(42)
@parameterized.parameters((16, 8, 200, 0.009), (31, 17, 100, 0.013),
(71, 41, 100, 0.045), (150, 100, 100, 0.129),
(300, 300, 100, 0.116), (600, 600, 50, 0.176))
def test_reference_match(self, n, top, runs, max_deviation):
"""Compares that new optimized method is close to reference method.
Runs two algorithms with same sets of input boxes and scores, and measures
deviation between returned sets of prunned boxes.
Read more about test results at ./g3doc/non_max_suppression.md
(*) Avoid flakiness with safe boundary (go/python-tips/048): deviation
between two sets is a positive number, which may vary from test to test.
Doing multiple runs expected to reduce average deviation variation following
LLN theorem. Therefore by having first test run we know upper deviation
bound which algorithm would not exceed until broken (in any feasible amount
of time in the future). Use of this safe boundary makes test non-flaky.
Args:
n: number of boxes and scores on input of the algorithm.
top: limit of output boxes count.
runs: for the statistical testing number of runs to performs to avoid
tests flakiness.
max_deviation: mean limit on deviation between optimized and reference
algorithms. Please read notes why this number may be set higher to avoid
flaky testing.
"""
deviation_rate = 0
min_union = 2*n
boxes = random_boxes([runs, n])
scores = tf.random.uniform(shape=[runs, n])
test = custom_layers.non_max_suppression_padded(boxes, scores, top)
for run in range(runs):
reference = tf.image.non_max_suppression(boxes[run], scores[run], top)
reference = {*reference.numpy().tolist()}
optimized = {*test[run].numpy().astype(int).tolist()} - {-1}
union_size = len(optimized | reference)
deviation_rate += len(optimized ^ reference) / union_size
min_union = min(min_union, union_size)
deviation_rate = deviation_rate / runs
# six sigma estimate via LLN theorem
safe_margin = 6 * (deviation_rate / np.sqrt(runs) + 1/(runs*min_union))
self.assertLess(
deviation_rate,
max_deviation,
msg='Deviation rate between optimized and reference implementations is '
'higher than expected. If you are tuning the test, recommended safe '
'deviation rate is '
f'{deviation_rate} + {safe_margin} = {deviation_rate + safe_margin}')
@parameterized.parameters(([16], 8), ([91, 150], 100), ([20, 20, 200], 10))
def test_sharded_match(self, shape: list[int], top: int):
boxes = random_boxes(shape)
scores = tf.random.uniform(shape=shape)
optimized = custom_layers.non_max_suppression_padded(boxes, scores, top)
reference = custom_layers._non_max_suppression_as_is(boxes, scores, top)
self.assertAllEqual(optimized, reference)
_sharded_nms = custom_layers.non_max_suppression_padded
_stright_nms = custom_layers._non_max_suppression_as_is
@parameterized.parameters(([16], 8, _sharded_nms, True),
([16], 8, _stright_nms, True),
([91, 150], 100, _sharded_nms, True),
([91, 150], 100, _stright_nms, False),
([20, 20, 200], 10, _sharded_nms, True),
([20, 20, 200], 10, _stright_nms, False))
def test_sharded_size(self, shape: list[int], top: int, algorithm,
fits_as_is: bool):
scores = tf.keras.Input(shape=shape, batch_size=1)
boxes = tf.keras.Input(shape=shape + [4], batch_size=1)
optimized = algorithm(boxes, scores, top)
model = tf.keras.Model(inputs=[boxes, scores], outputs=optimized)
max_size = _maximum_activation_size(model)
if fits_as_is:
# Sharding done or not needed.
self.assertLessEqual(max_size, custom_layers._RECOMMENDED_NMS_MEMORY)
else:
# Sharding needed.
self.assertGreater(max_size, custom_layers._RECOMMENDED_NMS_MEMORY)
def test_shard_tensors(self):
a: tf.Tensor = tf.constant([[0, 1, 2, 3, 4]])
b: tf.Tensor = tf.constant([[
[0, 1, 2, 3, 4],
[5, 6, 7, 8, 9],
[10, 11, 12, 13, 14],
[15, 16, 17, 18, 19],
[20, 21, 22, 23, 24],
]])
for i, (a_i, b_i) in enumerate(custom_layers.shard_tensors(1, 3, a, b)):
self.assertAllEqual(a_i, a[:, i * 3:i * 3 + 3])
self.assertAllEqual(b_i, b[:, i * 3:i * 3 + 3, :])
def test_top_k_sharded_fusion_arguments_validation(self):
# Input scores is not pair of agregation and shard.
self.assertRaises(ValueError, custom_layers.concat_and_top_k, 100,
tf.zeros(shape=[1000]))
# Input other values is not pairs of agregation and shard.
self.assertRaises(TypeError, custom_layers.concat_and_top_k, 100,
(None, tf.zeros(shape=[1000])), None,
tf.zeros(shape=[1000]))
# Insufficient rank to do top_k
self.assertRaises(IndexError, custom_layers.concat_and_top_k, 100,
(None, tf.constant(1.)))
@parameterized.parameters(0, 1, 2)
def test_top_k_sharded_fusion_vs_top_k_unsharded(self, axis: int):
r"""Tests `horizontal` sharding using shard_tensors and concat_and_top_k.
Will generate and test graph (on diagram 4 shards, in test 6 shards):
Input
-----
|
+-------+--------------------------------------------
| Split |----------------------- \
+-------+--- \ |
| \ | |
+-------+ +--------+ +-------+ +--------+ +-------+ +--------+ +-------+
| top k |-| concat |-| top k |-| concat |-| top k |-| concat |-| top k |
+-------+ +--------+ +-------+ +--------+ +-------+ +--------+ +-------+
|
Output
------
Args:
axis: test top_k axis (tensor rank will be axis + 1)
"""
sample: tf.Tensor = tf.random.uniform(
shape=axis * [1] + [10000], dtype=tf.float32)
top_1000_direct: tf.Tensor = tf.math.top_k(sample, 1000).values
top_1000_sharded: Optional[tf.Tensor] = None
for (piece,) in custom_layers.shard_tensors(axis, 1500, sample):
(top_1000_sharded,) = custom_layers.concat_and_top_k(
1000, (top_1000_sharded, piece))
self.assertAllEqual(top_1000_direct, top_1000_sharded)
if __name__ == '__main__':
tf.test.main()
official/vision/modeling/layers/detection_generator.py
浏览文件 @ 5938a718
......@@ -18,7 +18,7 @@ from typing import Any, Dict, List, Optional, Mapping, Sequence, Tuple
# Import libraries
import tensorflow as tf
from official.projects.edgetpu.vision.modeling import custom_layers
from official.vision.modeling.layers import edgetpu
from official.vision.ops import box_ops
from official.vision.ops import nms
from official.vision.ops import preprocess_ops
......@@ -428,7 +428,7 @@ def _generate_detections_v3(
boxes, scores, min_score_threshold=pre_nms_score_threshold)
# EdgeTPU-friendly class-wise NMS, -1 for invalid.
indices = custom_layers.non_max_suppression_padded(
indices = edgetpu.non_max_suppression_padded(
boxes,
scores,
max_num_detections,
......
official/vision/modeling/layers/edgetpu.py 0 → 100644
浏览文件 @ 5938a718
# Copyright 2022 The TensorFlow Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""EdgeTPU oriented layers and tools."""
from collections.abc import Iterable, Sequence
from typing import Optional
import numpy as np
import tensorflow as tf
_or = tf.maximum
_and = tf.minimum
_reduce_or = tf.reduce_max
def _tensor_sum_vectors(a, b):
a = tf.tile(tf.reshape(a, [1, -1, 1, a.shape[-1]]), [1, 1, a.shape[-1], 1])
b = tf.tile(tf.reshape(b, [1, -1, a.shape[-1], 1]), [1, 1, 1, a.shape[-1]])
return a + b
def _tensor_product_iou(boxes):
"""Computes pairwise IOU.
Reason to use 4-D tensors is to follow TPU compiler preference.
Args:
boxes: A 2-D float `Tensor` of shape `[num_boxes, 4]`.
Returns:
A 4-D float `Tensor` of shape `[1, 1, num_boxes, num_boxes]` containing
pairwise IOU.
"""
boxes_size = boxes.shape[-2]
# Code below will do frequent operands broadcasting.
# TPU compiler has (empirically) less issues broadcasting if
# - batch (first) dimension is 1. (Special consideration sharding)
# - there are 4 dimensions. (Standard traversal mapping)
# - last dimension is not 1. (Structure alignment)
tpu_friendly_shape = [1, -1, 1, boxes_size]
bottom, left, top, right = (
tf.reshape(side, tpu_friendly_shape) for side in tf.split(boxes, 4, -1))
height, width = top - bottom, right - left
area = height * width
area_sum = _tensor_sum_vectors(area, area)
bottom_pad, left_pad, top_pad, right_pad = (
tf.nn.relu(_tensor_sum_vectors(x, -x))
for x in (-bottom, -left, top, right))
height_pad, width_pad = bottom_pad + top_pad, left_pad + right_pad
intersection = tf.nn.relu(height - height_pad) * tf.nn.relu(width - width_pad)
union = area_sum - intersection
iou = tf.math.divide(intersection, union + _same(union))
return iou
def _greater(x):
"""Avoid non lowerable layers in boolean comparison.
Logical operation results in tensor of boolean type. However in serving such
a tensors cannot be cast to values because of NNAPI specs.
`tf.where` operation result in `select` instruction lowering, which not runs
well on all generations of edge-tpus.
Args:
x: any numeric tensor.
Returns:
tf.where(x > tf.zero_like(x), tf.one_like(x), tf.zero_like(x))
"""
x_clip = tf.minimum(tf.nn.relu(x), tf.constant(1, dtype=x.dtype))
return -tf.math.floor(-x_clip)
def _same(x):
"""Avoid non lowerable layers in boolean equality.
Logical operation results in tensor of boolean type. However in serving such
a tensors cannot be cast to values because of NNAPI specs.
`tf.where` operation result in `select` instruction lowering, which not runs
well on all generations of edge-tpus.
Args:
x: any numeric tensor.
Returns:
tf.where(x == tf.zero_like(x), tf.one_like(x), tf.zero_like(x))
"""
x_clip = tf.minimum(tf.abs(x), tf.constant(1, dtype=x.dtype))
return tf.constant(1, dtype=x.dtype) + tf.math.floor(-x_clip)
def shard_tensors(axis: int, block_size: int,
*tensors: tf.Tensor) -> Iterable[Sequence[tf.Tensor]]:
"""Consistently splits multiple tensors sharding-style.
Args:
axis: axis to be used to split tensors
block_size: block size to split tensors.
*tensors: list of tensors.
Returns:
List of shards, each shard has exactly one peace of each input tesnor.
Raises:
ValueError: if input tensors has different size of sharded dimension.
"""
for validate_axis in range(axis + 1):
consistent_length: int = tensors[0].shape[validate_axis]
for tensor in tensors:
if tensor.shape[validate_axis] != consistent_length:
raise ValueError('Inconsistent shapes in shard_tensors: first is '
f'{tensors[0].shape} and other is {tensor.shape}')
batch_size: int = tensors[0].shape[axis]
if block_size >= batch_size:
return [tensors]
else:
blocks = batch_size // block_size
remainder = batch_size % block_size
if remainder:
tensor_parts = []
for tensor in tensors:
shape: tf.TensorShape = tensor.shape
body: tf.Tensor = tf.slice(tensor, [0] * len(shape), [
size if i != axis else blocks * block_size
for i, size in enumerate(shape)
])
tail: tf.Tensor = tf.slice(tensor, [
0 if i != axis else (blocks * block_size)
for i, _ in enumerate(shape)
], [
size if i != axis else (size - blocks * block_size)
for i, size in enumerate(shape)
])
tensor_parts.append(tf.split(body, blocks, axis) + [tail])
return zip(*tensor_parts)
else:
return zip(*[tf.split(tensor, blocks, axis) for tensor in tensors])
# TODO(b/258007436): Number is based on existing compiler limitations while
# running bf16 NMS on edgetpu. Remove manual sharing when compiler issue will be
# fixed.
_RECOMMENDED_NMS_MEMORY = 360000
def non_max_suppression_padded(boxes: tf.Tensor,
scores: tf.Tensor,
output_size: int,
iou_threshold: float = 0.5) -> tf.Tensor:
"""Selects a subset of boxes which have highest score among IOU-similar boxes.
Prunes away boxes that have high intersection-over-union (IOU) overlap
with boxes having higher score. Boxes are supplied as `[y1, x1, y2, x2]`,
where `(y1, x1)` and `(y2, x2)` are the coordinates of any diagonal pair of
box corners. Note that this algorithm is agnostic to the coordinate system.
Thus translating or reflections of the coordinate system result in the same
boxes being selected by the algorithm. The output of this operation is a
set of integers indexing into the input collection of bounding boxes
representing the selected boxes.
Set will be returned padded on the right with `-1` values. The bounding
box coordinates corresponding to the selected indices can then be obtained
using the `tf.gather` operation. For example:
```python
selected_indices = vision.modeling.layers.non_max_suppression_padded(
boxes, scores, max_output_size, iou_threshold)
selected_boxes = tf.gather(boxes, selected_indices)
```
See following documetation for implementation details.
third_party/tensorflow_models/official/projects/edgetpu/vision/modeling/g3doc/non_max_suppression.md
Args:
boxes: A 2-D+ float `Tensor` of shape `[...batch_dims, num_boxes, 4]`.
scores: A 1-D+ float `Tensor` of shape `[...batch_dims, num_boxes]`
representing a single score corresponding to each box (each row of boxes).
output_size: A scalar integer `Tensor` representing the maximum number of
boxes to be selected by non-max suppression.
iou_threshold: A 0-D float tensor representing the threshold for deciding
whether boxes overlap too much with respect to IOU.
Returns:
A 1-D+ integer `Tensor` of shape `[...batch_dims, output_size]` representing
the selected indices from the boxes tensor and `-1` values for the padding.
"""
# Does partitioning job to help compiler converge with memory.
batch_shape = boxes.shape[:-2]
batch_size = np.prod(batch_shape, dtype=np.int32)
boxes_size, struct_size = boxes.shape[-2:]
boxes = tf.reshape(boxes, [batch_size, boxes_size, struct_size])
scores = tf.reshape(scores, [batch_size, boxes_size])
block = max(1, _RECOMMENDED_NMS_MEMORY // (boxes_size * boxes_size))
indices = []
for boxes_i, scores_i in shard_tensors(0, block, boxes, scores):
indices.append(
_non_max_suppression_as_is(boxes_i, scores_i, output_size,
iou_threshold))
indices = tf.concat(indices, axis=0)
return tf.reshape(indices, batch_shape + [output_size])
def _non_max_suppression_as_is(boxes: tf.Tensor,
scores: tf.Tensor,
output_size: int,
iou_threshold: float = 0.5) -> tf.Tensor:
"""Selects a subset of boxes which have highest score among IOU-similar boxes.
Args:
boxes: A 2-D+ float `Tensor` of shape `[...batch_dims, num_boxes, 4]`.
scores: A 1-D+ float `Tensor` of shape `[...batch_dims, num_boxes]`
representing a single score corresponding to each box (each row of boxes).
output_size: A scalar integer `Tensor` representing the maximum number of
boxes to be selected by non-max suppression.
iou_threshold: A 0-D float tensor representing the threshold for deciding
whether boxes overlap too much with respect to IOU.
Returns:
A 1-D+ integer `Tensor` of shape `[...batch_dims, output_size]` representing
the selected indices from the boxes tensor and `-1` values for the padding.
"""
batch_shape = boxes.shape[:-2]
batch_size = np.prod(batch_shape, dtype=np.int32)
boxes_size = boxes.shape[-2]
if boxes.shape[-1] != 4:
raise ValueError(f'Boxes shape ({boxes.shape}) last dimension must be 4 '
'to represent [y1, x1, y2, x2] boxes coordinates')
if scores.shape != boxes.shape[:-1]:
raise ValueError(f'Boxes shape ({boxes.shape}) and scores shape '
f'({scores.shape}) do not match.')
order = tf.range(boxes_size, dtype=tf.float32)
relative_order = _tensor_sum_vectors(order, -order)
relative_scores = _tensor_sum_vectors(scores, -scores)
similar = _greater(_tensor_product_iou(boxes) - iou_threshold)
worse = _greater(relative_scores)
same_later = _and(_same(relative_scores), _greater(relative_order))
similar_worse_or_same_later = _and(similar, _or(worse, same_later))
prunable = _reduce_or(similar_worse_or_same_later, axis=-1)
remaining = tf.constant(1.) - prunable
scores = tf.reshape(tf.exp(scores), [1, 1, batch_size, boxes_size])
remaining = tf.reshape(remaining, [1, 1, batch_size, boxes_size])
# top_k runs on TPU cores, let it happen, TPU tiles implementation is slower.
top_k = tf.math.top_k(scores * remaining, output_size)
indices = (
tf.cast(top_k.indices, top_k.values.dtype) * _greater(top_k.values) -
_same(top_k.values))
return tf.reshape(indices, batch_shape + [output_size])
def concat_and_top_k(
top_k: int, scores_pair: tuple[Optional[tf.Tensor], tf.Tensor],
*other_pairs: tuple[Optional[tf.Tensor], tf.Tensor]
) -> tuple[tf.Tensor, ...]:
"""Combines shards of top_k operation, when sharded along filtered dimension.
General idea is that sometimes top_k dimension is very large, while top_k is
moderately low. (Keep in mind sample of 15K pre-top_k dimension and 150 top_k)
In that case it is possible to break top_k input into groups significantly
larger than top_k and significatly lower than pre-top_l (Keep in mind 1500).
We do top_k over first 1500 elements, than join 150 remaining with new 1500
elements (1750 in total), repeat top_k. This function provides repeatedly used
method which will concat and top_k in that case.
For example with top_k = 2 and scores_pair = ([10, 6], [9, 8, 7]), output
scores will be [10, 9].
Other pairs are filtered using indexes generated from scores. This is a preaty
common case of filtering structure by its score.
For example with one extra pair of box per score:
top_k = 2
scores_pair = ([10, 6],
[9, 8, 7])
other_pairs = [([[0, 0, 10, 10], [0, 0, 6, 6]],
[[1, 1, 9, 9], [1, 1, 8, 8], [1, 1, 7, 7]])]
Output is:
([10, 9], [[0, 0, 10, 10], [1, 1, 9, 9]])
See also 'test_top_k_sharded_fusion' unit test with end to end example.
Args:
top_k: is top_k argument of sharded tf.math.top_k.
scores_pair: Tuple (<previous shards combination>, <additional shard>)
scores to be aggregated using top_k.
*other_pairs: Tuples (<previous shards combination>, <additional shard>)
other values to be aggregated using indexes of top_k scores.
Returns:
Tuple of scores based top_k aggregations with additional shards.
"""
scores, scores_shard = scores_pair
if other_pairs:
others, others_shard = zip(*other_pairs)
else:
others = others_shard = []
# Same as tf.rank, but avoiding tensor form for graph mode execution.
top_k_dim: int = len(scores_shard.shape) - 1
if scores is None:
# First shard becomes aggregation
scores = scores_shard
others = others_shard
else:
# Merge shard into aggregation
scores = tf.concat([scores, scores_shard], top_k_dim)
others = [
tf.concat([other, other_shard], top_k_dim)
for other, other_shard in zip(others, others_shard)
]
# When shards are uneven some will be smaller than requested top_k
if scores.shape[top_k_dim] > top_k:
scores, indices = tf.nn.top_k(scores, top_k)
others = [
tf.gather(other, indices, axis=top_k_dim, batch_dims=top_k_dim)
for other in others
]
return scores, *others
official/vision/modeling/layers/edgetpu_test.py 0 → 100644
浏览文件 @ 5938a718
# Copyright 2022 The TensorFlow Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""Tests EdgeTPU oriented layers and tools."""
from typing import Optional
from absl.testing import parameterized
import numpy as np
import tensorflow as tf
from official.vision.modeling.layers import edgetpu
def random_boxes(shape):
a = tf.random.uniform(shape=shape+[2])
b = tf.random.uniform(shape=shape+[2])
l = tf.minimum(a, b)
u = tf.maximum(a, b)
return tf.concat([l, u], axis=-1)
def _maximum_activation_size(model):
max_size = 0
for layer in model.layers:
outputs = layer.output
if not isinstance(outputs, list):
outputs = [outputs]
for output in outputs:
if hasattr(output, 'shape'):
size = np.prod(output.shape)
max_size = max(max_size, size)
print('Layer', size, output.shape, layer.name)
return max_size
class NonMaxSuppressionTest(parameterized.TestCase, tf.test.TestCase):
def setUp(self):
super().setUp()
tf.random.set_seed(42)
@parameterized.parameters((16, 8, 200, 0.009), (31, 17, 100, 0.013),
(71, 41, 100, 0.045), (150, 100, 100, 0.129),
(300, 300, 100, 0.116), (600, 600, 50, 0.176))
def test_reference_match(self, n, top, runs, max_deviation):
"""Compares that new optimized method is close to reference method.
Runs two algorithms with same sets of input boxes and scores, and measures
deviation between returned sets of prunned boxes.
Read more about test results at ./g3doc/non_max_suppression.md
(*) Avoid flakiness with safe boundary (go/python-tips/048): deviation
between two sets is a positive number, which may vary from test to test.
Doing multiple runs expected to reduce average deviation variation following
LLN theorem. Therefore by having first test run we know upper deviation
bound which algorithm would not exceed until broken (in any feasible amount
of time in the future). Use of this safe boundary makes test non-flaky.
Args:
n: number of boxes and scores on input of the algorithm.
top: limit of output boxes count.
runs: for the statistical testing number of runs to performs to avoid
tests flakiness.
max_deviation: mean limit on deviation between optimized and reference
algorithms. Please read notes why this number may be set higher to avoid
flaky testing.
"""
deviation_rate = 0
min_union = 2*n
boxes = random_boxes([runs, n])
scores = tf.random.uniform(shape=[runs, n])
test = edgetpu.non_max_suppression_padded(boxes, scores, top)
for run in range(runs):
reference = tf.image.non_max_suppression(boxes[run], scores[run], top)
reference = {*reference.numpy().tolist()}
optimized = {*test[run].numpy().astype(int).tolist()} - {-1}
union_size = len(optimized | reference)
deviation_rate += len(optimized ^ reference) / union_size
min_union = min(min_union, union_size)
deviation_rate = deviation_rate / runs
# six sigma estimate via LLN theorem
safe_margin = 6 * (deviation_rate / np.sqrt(runs) + 1/(runs*min_union))
self.assertLess(
deviation_rate,
max_deviation,
msg='Deviation rate between optimized and reference implementations is '
'higher than expected. If you are tuning the test, recommended safe '
'deviation rate is '
f'{deviation_rate} + {safe_margin} = {deviation_rate + safe_margin}')
@parameterized.parameters(([16], 8), ([91, 150], 100), ([20, 20, 200], 10))
def test_sharded_match(self, shape: list[int], top: int):
boxes = random_boxes(shape)
scores = tf.random.uniform(shape=shape)
optimized = edgetpu.non_max_suppression_padded(boxes, scores, top)
reference = edgetpu._non_max_suppression_as_is(boxes, scores, top)
self.assertAllEqual(optimized, reference)
_sharded_nms = edgetpu.non_max_suppression_padded
_stright_nms = edgetpu._non_max_suppression_as_is
@parameterized.parameters(([16], 8, _sharded_nms, True),
([16], 8, _stright_nms, True),
([91, 150], 100, _sharded_nms, True),
([91, 150], 100, _stright_nms, False),
([20, 20, 200], 10, _sharded_nms, True),
([20, 20, 200], 10, _stright_nms, False))
def test_sharded_size(self, shape: list[int], top: int, algorithm,
fits_as_is: bool):
scores = tf.keras.Input(shape=shape, batch_size=1)
boxes = tf.keras.Input(shape=shape + [4], batch_size=1)
optimized = algorithm(boxes, scores, top)
model = tf.keras.Model(inputs=[boxes, scores], outputs=optimized)
max_size = _maximum_activation_size(model)
if fits_as_is:
# Sharding done or not needed.
self.assertLessEqual(max_size, edgetpu._RECOMMENDED_NMS_MEMORY)
else:
# Sharding needed.
self.assertGreater(max_size, edgetpu._RECOMMENDED_NMS_MEMORY)
def test_shard_tensors(self):
a: tf.Tensor = tf.constant([[0, 1, 2, 3, 4]])
b: tf.Tensor = tf.constant([[
[0, 1, 2, 3, 4],
[5, 6, 7, 8, 9],
[10, 11, 12, 13, 14],
[15, 16, 17, 18, 19],
[20, 21, 22, 23, 24],
]])
for i, (a_i, b_i) in enumerate(edgetpu.shard_tensors(1, 3, a, b)):
self.assertAllEqual(a_i, a[:, i * 3:i * 3 + 3])
self.assertAllEqual(b_i, b[:, i * 3:i * 3 + 3, :])
def test_top_k_sharded_fusion_arguments_validation(self):
# Input scores is not pair of aggregation and shard.
self.assertRaises(ValueError, edgetpu.concat_and_top_k, 100,
tf.zeros(shape=[1000]))
# Input other values is not pairs of aggregation and shard.
self.assertRaises(TypeError, edgetpu.concat_and_top_k, 100,
(None, tf.zeros(shape=[1000])), None,
tf.zeros(shape=[1000]))
# Insufficient rank to do top_k
self.assertRaises(IndexError, edgetpu.concat_and_top_k, 100,
(None, tf.constant(1.)))
@parameterized.parameters(0, 1, 2)
def test_top_k_sharded_fusion_vs_top_k_unsharded(self, axis: int):
r"""Tests `horizontal` sharding using shard_tensors and concat_and_top_k.
Will generate and test graph (on diagram 4 shards, in test 6 shards):
Input
-----
|
+-------+--------------------------------------------
| Split |----------------------- \
+-------+--- \ |
| \ | |
+-------+ +--------+ +-------+ +--------+ +-------+ +--------+ +-------+
| top k |-| concat |-| top k |-| concat |-| top k |-| concat |-| top k |
+-------+ +--------+ +-------+ +--------+ +-------+ +--------+ +-------+
|
Output
------
Args:
axis: test top_k axis (tensor rank will be axis + 1)
"""
sample: tf.Tensor = tf.random.uniform(
shape=axis * [1] + [10000], dtype=tf.float32)
top_1000_direct: tf.Tensor = tf.math.top_k(sample, 1000).values
top_1000_sharded: Optional[tf.Tensor] = None
for (piece,) in edgetpu.shard_tensors(axis, 1500, sample):
(top_1000_sharded,) = edgetpu.concat_and_top_k(
1000, (top_1000_sharded, piece))
self.assertAllEqual(top_1000_direct, top_1000_sharded)
if __name__ == '__main__':
tf.test.main()
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或