# Copyright 2021 Huawei Technologies Co., Ltd
#
# 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.
# ============================================================================
"""SSD300 network based on MobileNetV2 backbone."""
import tinyms as ts
from .. import layers, primitives as P, Tensor
from ..vision.utils import ssd_default_boxes
def _make_divisible(v, divisor, min_value=None):
"""ensures that all layers have a channel number that is divisible by 8."""
if min_value is None:
min_value = divisor
new_v = max(min_value, int(v + divisor / 2) // divisor * divisor)
# Make sure that round down does not go down by more than 10%.
if new_v < 0.9 * v:
new_v += divisor
return new_v
def _conv2d(in_channel, out_channel, kernel_size=3, stride=1, pad_mod='same'):
return layers.Conv2d(in_channel, out_channel, kernel_size=kernel_size, stride=stride,
padding=0, pad_mode=pad_mod, has_bias=True)
def _bn(channel):
return layers.BatchNorm2d(channel, eps=1e-3, momentum=0.97,
gamma_init=1, beta_init=0, moving_mean_init=0, moving_var_init=1)
def _last_conv2d(in_channel, out_channel, kernel_size=3, stride=1, pad_mod='same', pad=0):
in_channels = in_channel
out_channels = in_channel
depthwise_conv = layers.Conv2d(in_channels, out_channels, kernel_size, stride, pad_mode=pad_mod,
padding=pad, group=in_channels)
conv = _conv2d(in_channel, out_channel, kernel_size=1)
return layers.SequentialLayer([depthwise_conv, _bn(in_channel), layers.ReLU6(), conv])
class ConvBNReLU(layers.Layer):
"""
Convolution/Depthwise fused with Batchnorm and ReLU block definition.
Args:
in_channels (int): Input channel.
out_channels (int): Output channel.
kernel_size (int): Input kernel size. Default: 3.
stride (int): Stride size for the first convolutional layer. Default: 1.
groups (int): channel group. Convolution is 1 while Depthwise is input channel. Default: 1.
Returns:
Tensor, output tensor.
Examples:
>>> ConvBNReLU(16, 256, kernel_size=1, stride=1, groups=1)
"""
def __init__(self, in_channels, out_channels, kernel_size=3, stride=1, groups=1):
super(ConvBNReLU, self).__init__()
padding = 0
if groups == 1:
conv = layers.Conv2d(in_channels, out_channels, kernel_size, stride,
pad_mode='same', padding=padding)
else:
conv = layers.Conv2d(in_channels, in_channels, kernel_size, stride,
pad_mode='same', padding=padding, group=in_channels)
self.features = layers.SequentialLayer([conv,
_bn(out_channels),
layers.ReLU6()])
def construct(self, x):
output = self.features(x)
return output
class InvertedResidual(layers.Layer):
def __init__(self, inp, oup, stride, expand_ratio, use_relu=False):
super(InvertedResidual, self).__init__()
assert stride in [1, 2]
hidden_dim = int(round(inp * expand_ratio))
self.use_res_connect = stride == 1 and inp == oup
residual_layers = []
if expand_ratio != 1:
residual_layers.append(ConvBNReLU(inp, hidden_dim, kernel_size=1))
residual_layers.extend([
ConvBNReLU(hidden_dim, hidden_dim, stride=stride, groups=hidden_dim),
layers.Conv2d(hidden_dim, oup, kernel_size=1, stride=1, has_bias=False),
_bn(oup),
])
self.conv = layers.SequentialLayer(residual_layers)
self.use_relu = use_relu
self.relu = layers.ReLU6()
def construct(self, x):
identity = x
x = self.conv(x)
if self.use_res_connect:
return P.tensor_add(identity, x)
if self.use_relu:
x = self.relu(x)
return x
class FlattenConcat(layers.Layer):
"""
Concatenate predictions into a single tensor.
Args:
num_ssd_boxes (int): number of ssd boxes. Default is 1917.
Returns:
Tensor, flatten predictions.
"""
def __init__(self, num_ssd_boxes=1917):
super(FlattenConcat, self).__init__()
self.num_ssd_boxes = num_ssd_boxes
self.concat = P.Concat(axis=1)
self.transpose = P.Transpose()
def construct(self, inputs):
output = ()
batch_size = P.shape(inputs[0])[0]
for x in inputs:
x = self.transpose(x, (0, 2, 3, 1))
output += (P.reshape(x, (batch_size, -1)),)
res = self.concat(output)
return P.reshape(res, (batch_size, self.num_ssd_boxes, -1))
class MultiBox(layers.Layer):
"""
Multibox conv layers. Each multibox layer contains class conf scores
and localization predictions.
Args:
class_num (int): number of classes. Default: 21.
Returns:
Tensor, localization predictions.
Tensor, class conf scores.
"""
def __init__(self, class_num=21):
super(MultiBox, self).__init__()
out_channels = [576, 1280, 512, 256, 256, 128]
num_default = [3, 6, 6, 6, 6, 6]
loc_layers = []
cls_layers = []
for k, out_channel in enumerate(out_channels):
loc_layers += [_last_conv2d(out_channel, 4 * num_default[k],
kernel_size=3, stride=1, pad_mod='same', pad=0)]
cls_layers += [_last_conv2d(out_channel, class_num * num_default[k],
kernel_size=3, stride=1, pad_mod='same', pad=0)]
self.multi_loc_layers = layers.LayerList(loc_layers)
self.multi_cls_layers = layers.LayerList(cls_layers)
self.flatten_concat = FlattenConcat()
def construct(self, inputs):
loc_outputs = ()
cls_outputs = ()
for i in range(len(self.multi_loc_layers)):
loc_outputs += (self.multi_loc_layers[i](inputs[i]),)
cls_outputs += (self.multi_cls_layers[i](inputs[i]),)
return self.flatten_concat(loc_outputs), self.flatten_concat(cls_outputs)
class SSDWithMobileNetV2(layers.Layer):
"""
MobileNetV2 backbone.
Args:
width_mult (float): Channels multiplier for round to 8/16 and others. Default: 1.0.
round_nearest (int): Channel round to. Default: 8.
input_channel (int): Input channel. Default: 32.
last_channel (int): The channel of last layer. Default: 1280.
Returns:
Tensor, output feature maps.
Tensor, output tensor.
"""
def __init__(self, width_mult=1.0, round_nearest=8, input_channel=32, last_channel=1280):
super(SSDWithMobileNetV2, self).__init__()
# setting of inverted residual blocks
cfg = [
# t, c, n, s
[1, 16, 1, 1],
[6, 24, 2, 2],
[6, 32, 3, 2],
[6, 64, 4, 2],
[6, 96, 3, 1],
[6, 160, 3, 2],
[6, 320, 1, 1],
]
# building first layer
input_channel = _make_divisible(input_channel * width_mult, round_nearest)
self.last_channel = _make_divisible(last_channel * max(1.0, width_mult), round_nearest)
backbone_layers = [ConvBNReLU(3, input_channel, stride=2)]
# building inverted residual blocks
layer_index = 0
for t, c, n, s in cfg:
output_channel = _make_divisible(c * width_mult, round_nearest)
for i in range(n):
if layer_index == 13:
hidden_dim = int(round(input_channel * t))
self.expand_layer_conv_13 = ConvBNReLU(input_channel, hidden_dim, kernel_size=1)
stride = s if i == 0 else 1
backbone_layers.append(InvertedResidual(input_channel, output_channel, stride, expand_ratio=t))
input_channel = output_channel
layer_index += 1
# building last several layers
backbone_layers.append(ConvBNReLU(input_channel, self.last_channel, kernel_size=1))
self.features_1 = layers.SequentialLayer(backbone_layers[:14])
self.features_2 = layers.SequentialLayer(backbone_layers[14:])
def construct(self, x):
out = self.features_1(x)
expand_layer_conv_13 = self.expand_layer_conv_13(out)
out = self.features_2(out)
return expand_layer_conv_13, out
[docs]class SSD300(layers.Layer):
"""
SSD300 Network. Default backbone is MobileNetV2.
Args:
backbone (layers.Layer): backbone of ssd300 model.
class_num (int): number of classes. Default: 21.
is_training (bool): Specify if in training step. Default: True.
Returns:
Tensor, localization predictions.
Tensor, class conf scores.
"""
def __init__(self, backbone, class_num=21, is_training=True):
super(SSD300, self).__init__()
self.backbone = backbone
in_channels = [256, 576, 1280, 512, 256, 256]
out_channels = [576, 1280, 512, 256, 256, 128]
ratios = [0.2, 0.2, 0.2, 0.25, 0.5, 0.25]
strides = [1, 1, 2, 2, 2, 2]
residual_list = []
for i in range(2, len(in_channels)):
residual = InvertedResidual(in_channels[i], out_channels[i], stride=strides[i],
expand_ratio=ratios[i], use_relu=True)
residual_list.append(residual)
self.multi_residual = layers.LayerList(residual_list)
self.multi_box = MultiBox(class_num=class_num)
self.is_training = is_training
self.activation = P.Sigmoid()
def construct(self, x):
layer_out_13, output = self.backbone(x)
multi_feature = (layer_out_13, output)
feature = output
for residual in self.multi_residual:
feature = residual(feature)
multi_feature += (feature,)
pred_loc, pred_label = self.multi_box(multi_feature)
if not self.is_training:
pred_label = self.activation(pred_label)
pred_loc = P.cast(pred_loc, ts.float32)
pred_label = P.cast(pred_label, ts.float32)
return pred_loc, pred_label
def ssd300_mobilenetv2(class_num=21, is_training=True):
return SSD300(SSDWithMobileNetV2(), class_num=class_num, is_training=is_training)
class SSDInferWithDecoder(layers.Layer):
"""
SSD300 infer wrapper to decode the bbox locations.
Args:
network (layers.Layer): the origin ssd300 infer network without bbox decoder.
Returns:
Tensor, the locations for bbox after decoder representing (y0, x0, y1, x1)
Tensor, the prediction labels.
"""
def __init__(self, network):
super(SSDInferWithDecoder, self).__init__()
self.network = network
self.default_boxes = Tensor(ssd_default_boxes, dtype=ts.float32)
prior_scaling_tensor = Tensor((0.1, 0.2), dtype=ts.float32)
self.prior_scaling_xy = prior_scaling_tensor[0]
self.prior_scaling_wh = prior_scaling_tensor[1]
def construct(self, x):
pred_loc, pred_label = self.network(x)
default_bbox_xy = self.default_boxes[..., :2]
default_bbox_wh = self.default_boxes[..., 2:]
pred_xy = pred_loc[..., :2] * self.prior_scaling_xy * default_bbox_wh + default_bbox_xy
pred_wh = P.Exp()(pred_loc[..., 2:] * self.prior_scaling_wh) * default_bbox_wh
pred_xy_0 = pred_xy - pred_wh / 2.0
pred_xy_1 = pred_xy + pred_wh / 2.0
pred_xy = P.Concat(-1)((pred_xy_0, pred_xy_1))
pred_xy = P.Maximum()(pred_xy, 0)
pred_xy = P.Minimum()(pred_xy, 1)
return pred_xy, pred_label
def ssd300_infer(class_num=21):
net = SSD300(SSDWithMobileNetV2(), class_num=class_num, is_training=False)
return SSDInferWithDecoder(net)