wehub-resource-sync
272 lines
7.6 KiB
Python
272 lines
7.6 KiB
Python
# Copyright (c) 2020 PaddlePaddle 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.
|
|
|
|
import itertools
|
|
import unittest
|
|
|
|
import numpy as np
|
|
|
|
import paddle
|
|
from paddle import base
|
|
|
|
paddle.enable_static()
|
|
|
|
|
|
def corr(
|
|
x_1,
|
|
x_2,
|
|
pad_size=4,
|
|
kernel_size=1,
|
|
max_displacement=4,
|
|
stride1=1,
|
|
stride2=1,
|
|
corr_multiply=1,
|
|
):
|
|
K = kernel_size
|
|
|
|
rinput1 = np.pad(
|
|
x_1,
|
|
((0, 0), (0, 0), (pad_size, pad_size), (pad_size, pad_size)),
|
|
mode='constant',
|
|
)
|
|
rinput2 = np.pad(
|
|
x_2,
|
|
((0, 0), (0, 0), (pad_size, pad_size), (pad_size, pad_size)),
|
|
mode='constant',
|
|
)
|
|
rinput1 = np.transpose(rinput1, (0, 2, 3, 1))
|
|
rinput2 = np.transpose(rinput2, (0, 2, 3, 1))
|
|
B = int(rinput1.shape[0])
|
|
H = int(x_1.shape[2])
|
|
W = int(x_2.shape[3])
|
|
d = max_displacement
|
|
D = 2 * d + 1
|
|
output = np.zeros((B, D * D, H, W), dtype=np.float32)
|
|
|
|
for b, i, j, k, l in itertools.product(
|
|
range(B),
|
|
range(H),
|
|
range(W),
|
|
range(-d, d + 1),
|
|
range(-d, d + 1),
|
|
):
|
|
x1_index = i + pad_size
|
|
y1_index = j + pad_size
|
|
x2_index = x1_index + k
|
|
y2_index = y1_index + l
|
|
output[b, l + d + D * (k + d), i, j] = np.mean(
|
|
rinput1[
|
|
b,
|
|
x1_index : x1_index + K,
|
|
y1_index : y1_index + K,
|
|
]
|
|
* rinput2[
|
|
b,
|
|
x2_index : x2_index + K,
|
|
y2_index : y2_index + K,
|
|
]
|
|
)
|
|
|
|
return output
|
|
|
|
|
|
class TestCorrelationOp(unittest.TestCase):
|
|
def test_check_output(self):
|
|
if not base.core.is_compiled_with_cuda():
|
|
return
|
|
np.random.seed(13)
|
|
np.set_printoptions(threshold=np.inf)
|
|
x_shape = (2, 10, 3, 3)
|
|
x_type = 'float32'
|
|
x1 = paddle.static.data(
|
|
name='x1',
|
|
shape=x_shape,
|
|
dtype=x_type,
|
|
)
|
|
x1.desc.set_need_check_feed(False)
|
|
x1.stop_gradient = False
|
|
x2 = paddle.static.data(
|
|
name='x2',
|
|
shape=x_shape,
|
|
dtype=x_type,
|
|
)
|
|
x2.desc.set_need_check_feed(False)
|
|
x2.stop_gradient = False
|
|
|
|
x1_np = np.random.randn(2, 3, 4, 5).astype(x_type)
|
|
x2_np = np.random.randn(2, 3, 4, 5).astype(x_type)
|
|
out_np = corr(
|
|
x1_np,
|
|
x2_np,
|
|
pad_size=4,
|
|
kernel_size=1,
|
|
max_displacement=4,
|
|
stride1=1,
|
|
stride2=1,
|
|
)
|
|
|
|
out = paddle.incubate.layers.correlation(
|
|
x1,
|
|
x2,
|
|
pad_size=4,
|
|
kernel_size=1,
|
|
max_displacement=4,
|
|
stride1=1,
|
|
stride2=1,
|
|
)
|
|
|
|
loss = paddle.mean(out)
|
|
optimizer = paddle.optimizer.Momentum(0.0001, 0.9)
|
|
optimizer.minimize(loss)
|
|
|
|
place = base.CUDAPlace(0)
|
|
exe = base.Executor(place)
|
|
res = exe.run(feed={'x1': x1_np, 'x2': x2_np}, fetch_list=[out, loss])
|
|
|
|
np.testing.assert_allclose(res[0], out_np, rtol=1e-05, atol=1e-8)
|
|
|
|
|
|
class Net(paddle.nn.Layer):
|
|
def __init__(self, name_scope):
|
|
super().__init__(name_scope)
|
|
|
|
def forward(self, x1, x2):
|
|
y = paddle.incubate.layers.correlation(
|
|
x1,
|
|
x2,
|
|
pad_size=4,
|
|
kernel_size=1,
|
|
max_displacement=4,
|
|
stride1=1,
|
|
stride2=1,
|
|
)
|
|
return y
|
|
|
|
|
|
class TestCorrelationOpDyGraph(unittest.TestCase):
|
|
def test_check_output(self):
|
|
if not base.core.is_compiled_with_cuda():
|
|
return
|
|
np.random.seed(13)
|
|
np.set_printoptions(threshold=np.inf)
|
|
x_shape = (2, 10, 3, 3)
|
|
x_type = 'float32'
|
|
place = base.CUDAPlace(0)
|
|
with base.dygraph.guard(place):
|
|
x1_np = np.random.randn(2, 3, 4, 5).astype(x_type)
|
|
x2_np = np.random.randn(2, 3, 4, 5).astype(x_type)
|
|
out_np = corr(
|
|
x1_np,
|
|
x2_np,
|
|
pad_size=4,
|
|
kernel_size=1,
|
|
max_displacement=4,
|
|
stride1=1,
|
|
stride2=1,
|
|
)
|
|
|
|
x1 = paddle.to_tensor(x1_np)
|
|
x2 = paddle.to_tensor(x2_np)
|
|
corr_pd = Net('corr_pd')
|
|
y = corr_pd(x1, x2)
|
|
out = y.numpy()
|
|
np.testing.assert_allclose(out, out_np, rtol=1e-05, atol=1e-8)
|
|
|
|
def test_check_grad_numeric(self):
|
|
if not base.core.is_compiled_with_cuda():
|
|
return
|
|
np.random.seed(13)
|
|
eps = 1e-3
|
|
x_type = 'float32'
|
|
place = base.CUDAPlace(0)
|
|
|
|
with base.dygraph.guard(place):
|
|
x1_np = np.random.randn(2, 3, 4, 5).astype(x_type)
|
|
x2_np = np.random.randn(2, 3, 4, 5).astype(x_type)
|
|
|
|
x1 = paddle.to_tensor(x1_np, stop_gradient=False)
|
|
x2 = paddle.to_tensor(x2_np, stop_gradient=False)
|
|
corr_pd = Net('corr_pd')
|
|
y = corr_pd(x1, x2)
|
|
|
|
grad_y = np.random.randn(*y.shape).astype(x_type)
|
|
|
|
dx1, dx2 = paddle.autograd.grad(
|
|
outputs=y,
|
|
inputs=[x1, x2],
|
|
grad_outputs=paddle.to_tensor(grad_y),
|
|
)
|
|
|
|
dx1_num = np.zeros_like(x1_np)
|
|
for idx in np.ndindex(*x1_np.shape):
|
|
x1_pos = x1_np.copy()
|
|
x1_neg = x1_np.copy()
|
|
x1_pos[idx] += eps
|
|
x1_neg[idx] -= eps
|
|
out_pos = corr(
|
|
x1_pos,
|
|
x2_np,
|
|
pad_size=4,
|
|
kernel_size=1,
|
|
max_displacement=4,
|
|
stride1=1,
|
|
stride2=1,
|
|
)
|
|
out_neg = corr(
|
|
x1_neg,
|
|
x2_np,
|
|
pad_size=4,
|
|
kernel_size=1,
|
|
max_displacement=4,
|
|
stride1=1,
|
|
stride2=1,
|
|
)
|
|
dx1_num[idx] = np.sum((out_pos - out_neg) * grad_y) / (2 * eps)
|
|
|
|
dx2_num = np.zeros_like(x2_np)
|
|
for idx in np.ndindex(*x2_np.shape):
|
|
x2_pos = x2_np.copy()
|
|
x2_neg = x2_np.copy()
|
|
x2_pos[idx] += eps
|
|
x2_neg[idx] -= eps
|
|
out_pos = corr(
|
|
x1_np,
|
|
x2_pos,
|
|
pad_size=4,
|
|
kernel_size=1,
|
|
max_displacement=4,
|
|
stride1=1,
|
|
stride2=1,
|
|
)
|
|
out_neg = corr(
|
|
x1_np,
|
|
x2_neg,
|
|
pad_size=4,
|
|
kernel_size=1,
|
|
max_displacement=4,
|
|
stride1=1,
|
|
stride2=1,
|
|
)
|
|
dx2_num[idx] = np.sum((out_pos - out_neg) * grad_y) / (2 * eps)
|
|
np.testing.assert_allclose(
|
|
dx1.numpy(), dx1_num, rtol=1e-3, atol=1e-3
|
|
)
|
|
np.testing.assert_allclose(
|
|
dx2.numpy(), dx2_num, rtol=1e-3, atol=1e-3
|
|
)
|
|
|
|
|
|
if __name__ == '__main__':
|
|
unittest.main()
|