def test_1(self):
input = np.random.random([3, 4, 5, 6]).astype("float64")
minus_1 = fluid.layers.fill_constant([1], "int32", -1)
minus_3 = fluid.layers.fill_constant([1], "int32", -3)
starts = fluid.layers.data(
name='starts', shape=[3], dtype='int32', append_batch_size=False)
ends = fluid.layers.data(
name='ends', shape=[3], dtype='int32', append_batch_size=False)
strides = fluid.layers.data(
name='strides', shape=[3], dtype='int32', append_batch_size=False)
x = fluid.layers.data(
name="x",
shape=[3, 4, 5, 6],
append_batch_size=False,
dtype="float64")
out_1 = paddle.strided_slice(
x,
axes=[0, 1, 2],
starts=[-3, 0, 2],
ends=[3, 100, -1],
strides=[1, 1, 1])
out_2 = paddle.strided_slice(
x,
axes=[0, 1, 3],
starts=[minus_3, 0, 2],
ends=[3, 100, -1],
strides=[1, 1, 1])
out_3 = paddle.strided_slice(
x,
axes=[0, 1, 3],
starts=[minus_3, 0, 2],
ends=[3, 100, minus_1],
strides=[1, 1, 1])
out_4 = paddle.strided_slice(
x, axes=[0, 1, 2], starts=starts, ends=ends, strides=strides)
out_5 = x[-3:3, 0:100:2, -1:2:-1]
out_6 = x[minus_3:3:1, 0:100:2, :, minus_1:2:minus_1]
out_7 = x[minus_1, 0:100:2, :, -1:2:-1]
exe = fluid.Executor(place=fluid.CPUPlace())
res_1, res_2, res_3, res_4, res_5, res_6, res_7 = exe.run(
fluid.default_main_program(),
feed={
"x": input,
'starts': np.array([-3, 0, 2]).astype("int32"),
'ends': np.array([3, 2147483648, -1]).astype("int64"),
'strides': np.array([1, 1, 1]).astype("int32")
},
fetch_list=[out_1, out_2, out_3, out_4, out_5, out_6, out_7])
assert np.array_equal(res_1, input[-3:3, 0:100, 2:-1, :])
assert np.array_equal(res_2, input[-3:3, 0:100, :, 2:-1])
assert np.array_equal(res_3, input[-3:3, 0:100, :, 2:-1])
assert np.array_equal(res_4, input[-3:3, 0:100, 2:-1, :])
assert np.array_equal(res_5, input[-3:3, 0:100:2, -1:2:-1, :])
assert np.array_equal(res_6, input[-3:3, 0:100:2, :, -1:2:-1])
assert np.array_equal(res_7, input[-1, 0:100:2, :, -1:2:-1])
def slice_select_prim2orig(op, x):
return paddle.strided_slice(
x,
axes=op.attr('axis'),
starts=op.attr('starts'),
ends=op.attr('ends'),
strides=op.attr('strides'))
def test_dygraph_op(self):
x = paddle.zeros(shape=[3, 4, 5, 6], dtype="float32")
axes = [1, 2, 3]
starts = [-3, 0, 2]
ends = [3, 2, 4]
strides_1 = [1, 1, 1]
sliced_1 = paddle.strided_slice(
x, axes=axes, starts=starts, ends=ends, strides=strides_1)
assert sliced_1.shape == (3, 2, 2, 2)
def test_compare_paddle_strided_slice_with_numpy(self):
paddle.disable_static()
array = np.arange(5)
pt = paddle.to_tensor(array)
s1 = 3
e1 = 1
stride1 = -2
sl = paddle.strided_slice(
pt, axes=[0, ], starts=[s1, ], ends=[e1, ], strides=[stride1, ])
self.assertTrue(array[s1:e1:stride1], sl)
array = np.arange(6 * 6).reshape((6, 6))
pt = paddle.to_tensor(array)
s2 = [8, -1]
e2 = [1, -5]
stride2 = [-2, -3]
sl = paddle.strided_slice(
pt, axes=[0, 1], starts=s2, ends=e2, strides=stride2)
self.assertTrue(
np.array_equal(sl.numpy(), array[s2[0]:e2[0]:stride2[0], s2[1]:e2[
1]:stride2[1]]))
array = np.arange(6 * 7 * 8).reshape((6, 7, 8))
pt = paddle.to_tensor(array)
s2 = [7, -1]
e2 = [2, -5]
stride2 = [-2, -3]
sl = paddle.strided_slice(
pt, axes=[0, 2], starts=s2, ends=e2, strides=stride2)
array_slice = array[s2[0]:e2[0]:stride2[0], ::, s2[1]:e2[1]:stride2[1]]
self.assertTrue(
np.array_equal(sl.numpy(), array_slice),
msg="paddle.strided_slice:\n {} \n numpy slice:\n{}".format(
sl.numpy(), array_slice))
def forward(self, inputs):
"""
forward
"""
axes = self.config['axes']
starts = self.config['starts']
ends = self.config['ends']
# strides = self.config['strides']
# strides = [1, paddle.to_tensor(1).astype('int32'), 1]
# strides = [1, paddle.to_tensor(1, dtype='int32'), 1]
strides = [1, paddle.to_tensor(np.array(1).astype("int32")), 1]
x = paddle.strided_slice(inputs,
axes=axes,
starts=starts,
ends=ends,
strides=strides)
return x
def forward(self, inputs):
"""
forward
"""
axes = self.config['axes']
starts = self.config['starts']
ends = self.config['ends']
strides = self.config['strides']
if self.config['isStartsTensor']:
starts = paddle.to_tensor(np.array(starts).astype('int32'))
if self.config['isEndsTensor']:
ends = paddle.to_tensor(np.array(ends).astype('int32'))
if self.config['isStridesTensor']:
strides = paddle.to_tensor(np.array(strides).astype('int32'))
x = paddle.strided_slice(inputs,
axes=axes,
starts=starts,
ends=ends,
strides=strides)
return x
def forward(self):
input = self.input('Input', 0)
weight = self.input('Filter', 0)
mask = self.input('Mask', 0)
offset = self.input('Offset', 0)
input = layers.pad2d(input, self.padding)
input_shape = paddle.shape(input)
if self.padded_x_h < 0 or self.padded_x_w < 0:
self.padded_x_h = input_shape[2]
self.padded_x_w = input_shape[3]
offset_x = paddle.strided_slice(offset,
axes=[1],
starts=[0],
ends=[self.offset_channel],
strides=[2])
offset_y = paddle.strided_slice(offset,
axes=[1],
starts=[1],
ends=[self.offset_channel],
strides=[2])
offset = paddle.concat([offset_x, offset_y], axis=1)
offset_shape = paddle.shape(offset)
offset_h = offset_shape[2]
offset_w = offset_shape[3]
coordinate = self.get_offset_coordinate(offset, 'float32',
offset_shape)
coordinate = coordinate.transpose((0, 2, 3, 1))
coord_lt, coord_rb, coord_lb, coord_rt = self.get_bilinear_corner_coordinate(
coordinate, self.padded_x_h, self.padded_x_w)
# clip coordinate
coordinate = paddle.concat([
paddle.clip(coordinate[:, :, :, :self.N], 0, self.padded_x_h - 1),
paddle.clip(coordinate[:, :, :, self.N:], 0, self.padded_x_w - 1)
],
axis=-1)
cof_lt, cof_rb, cof_lb, cof_rt = self.get_bilinear_coefficient(
coord_lt, coord_rb, coord_lb, coord_rt, coordinate)
feature_lt = self.get_feature_by_coordinate(input, coord_lt, offset_h,
offset_w, self.padded_x_w)
feature_rb = self.get_feature_by_coordinate(input, coord_rb, offset_h,
offset_w, self.padded_x_w)
feature_lb = self.get_feature_by_coordinate(input, coord_lb, offset_h,
offset_w, self.padded_x_w)
feature_rt = self.get_feature_by_coordinate(input, coord_rt, offset_h,
offset_w, self.padded_x_w)
feature_after_deformation = paddle.unsqueeze(cof_lt, 1) * feature_lt + \
paddle.unsqueeze(cof_rb, 1) * feature_rb + \
paddle.unsqueeze(cof_lb, 1) * feature_lb + \
paddle.unsqueeze(cof_rt, 1) * feature_rt
# modulation
if mask is not None:
mask = paddle.transpose(mask, (0, 2, 3, 1))
mask = paddle.unsqueeze(mask, 1)
mask = paddle.tile(mask, [1, self.in_channel, 1, 1, 1])
feature_after_deformation *= mask
feature_after_deformation = self.reshape_feature(
feature_after_deformation, offset_h, offset_w)
out = paddle.nn.functional.conv2d(feature_after_deformation,
weight,
stride=self.kernel_size,
groups=self.groups)
return {'Output': [out]}