PaddleClas/ppcls/arch/backbone/base/dbb/dbb_transforms.py

74 lines
2.8 KiB
Python

# copyright (c) 2023 PaddlePaddle Authors. All Rights Reserve.
#
# 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.
# reference: https://arxiv.org/abs/2103.13425, https://github.com/DingXiaoH/DiverseBranchBlock
import numpy as np
import paddle
import paddle.nn.functional as F
def transI_fusebn(kernel, bn):
gamma = bn.weight
std = (bn._variance + bn._epsilon).sqrt()
return kernel * (
(gamma / std).reshape([-1, 1, 1, 1])), bn.bias - bn._mean * gamma / std
def transII_addbranch(kernels, biases):
return sum(kernels), sum(biases)
def transIII_1x1_kxk(k1, b1, k2, b2, groups):
if groups == 1:
k = F.conv2d(k2, k1.transpose([1, 0, 2, 3]))
b_hat = (k2 * b1.reshape([1, -1, 1, 1])).sum((1, 2, 3))
else:
k_slices = []
b_slices = []
k1_T = k1.transpose([1, 0, 2, 3])
k1_group_width = k1.shape[0] // groups
k2_group_width = k2.shape[0] // groups
for g in range(groups):
k1_T_slice = k1_T[:, g * k1_group_width:(g + 1) *
k1_group_width, :, :]
k2_slice = k2[g * k2_group_width:(g + 1) * k2_group_width, :, :, :]
k_slices.append(F.conv2d(k2_slice, k1_T_slice))
b_slices.append((k2_slice * b1[g * k1_group_width:(
g + 1) * k1_group_width].reshape([1, -1, 1, 1])).sum((1, 2, 3
)))
k, b_hat = transIV_depthconcat(k_slices, b_slices)
return k, b_hat + b2
def transIV_depthconcat(kernels, biases):
return paddle.cat(kernels, axis=0), paddle.cat(biases)
def transV_avg(channels, kernel_size, groups):
input_dim = channels // groups
k = paddle.zeros((channels, input_dim, kernel_size, kernel_size))
k[np.arange(channels), np.tile(np.arange(input_dim),
groups), :, :] = 1.0 / kernel_size**2
return k
# This has not been tested with non-square kernels (kernel.shape[2] != kernel.shape[3]) nor even-size kernels
def transVI_multiscale(kernel, target_kernel_size):
H_pixels_to_pad = (target_kernel_size - kernel.shape[2]) // 2
W_pixels_to_pad = (target_kernel_size - kernel.shape[3]) // 2
return F.pad(
kernel,
[H_pixels_to_pad, H_pixels_to_pad, W_pixels_to_pad, W_pixels_to_pad])