# Copyright (c) 2020 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. import math import paddle.fluid as fluid from paddle.fluid.param_attr import ParamAttr from paddle.fluid.regularizer import L2Decay from paddle.fluid.dygraph.nn import Conv2D, BatchNorm, Pool2D, Linear from paddle.incubate.hapi.model import Model from paddle.incubate.hapi.download import get_weights_path_from_url __all__ = ['DarkNet', 'darknet53'] # {num_layers: (url, md5)} model_urls = { 'darknet53': ('https://paddle-hapi.bj.bcebos.com/models/darknet53.pdparams', 'ca506a90e2efecb9a2093f8ada808708') } class ConvBNLayer(fluid.dygraph.Layer): def __init__(self, ch_in, ch_out, filter_size=3, stride=1, groups=1, padding=0, act="leaky"): super(ConvBNLayer, self).__init__() self.conv = Conv2D( num_channels=ch_in, num_filters=ch_out, filter_size=filter_size, stride=stride, padding=padding, groups=groups, param_attr=ParamAttr( initializer=fluid.initializer.Normal(0., 0.02)), bias_attr=False, act=None) self.batch_norm = BatchNorm( num_channels=ch_out, param_attr=ParamAttr( initializer=fluid.initializer.Normal(0., 0.02), regularizer=L2Decay(0.)), bias_attr=ParamAttr( initializer=fluid.initializer.Constant(0.0), regularizer=L2Decay(0.))) self.act = act def forward(self, inputs): out = self.conv(inputs) out = self.batch_norm(out) if self.act == 'leaky': out = fluid.layers.leaky_relu(x=out, alpha=0.1) return out class DownSample(fluid.dygraph.Layer): def __init__(self, ch_in, ch_out, filter_size=3, stride=2, padding=1): super(DownSample, self).__init__() self.conv_bn_layer = ConvBNLayer( ch_in=ch_in, ch_out=ch_out, filter_size=filter_size, stride=stride, padding=padding) self.ch_out = ch_out def forward(self, inputs): out = self.conv_bn_layer(inputs) return out class BasicBlock(fluid.dygraph.Layer): def __init__(self, ch_in, ch_out): super(BasicBlock, self).__init__() self.conv1 = ConvBNLayer( ch_in=ch_in, ch_out=ch_out, filter_size=1, stride=1, padding=0) self.conv2 = ConvBNLayer( ch_in=ch_out, ch_out=ch_out * 2, filter_size=3, stride=1, padding=1) def forward(self, inputs): conv1 = self.conv1(inputs) conv2 = self.conv2(conv1) out = fluid.layers.elementwise_add(x=inputs, y=conv2, act=None) return out class LayerWarp(fluid.dygraph.Layer): def __init__(self, ch_in, ch_out, count): super(LayerWarp, self).__init__() self.basicblock0 = BasicBlock(ch_in, ch_out) self.res_out_list = [] for i in range(1, count): res_out = self.add_sublayer("basic_block_%d" % (i), BasicBlock(ch_out * 2, ch_out)) self.res_out_list.append(res_out) self.ch_out = ch_out def forward(self, inputs): y = self.basicblock0(inputs) for basic_block_i in self.res_out_list: y = basic_block_i(y) return y DarkNet_cfg = {53: ([1, 2, 8, 8, 4])} class DarkNet(Model): """DarkNet model from `"YOLOv3: An Incremental Improvement" `_ Args: num_layers (int): layer number of DarkNet, only 53 supported currently, default: 53. num_classes (int): output dim of last fc layer. If num_classes <=0, last fc layer will not be defined. Default: 1000. with_pool (bool): use pool before the last fc layer or not. Default: True. classifier_activation (str): activation for the last fc layer. Default: 'softmax'. """ def __init__(self, num_layers=53, num_classes=1000, with_pool=True, classifier_activation='softmax'): super(DarkNet, self).__init__() assert num_layers in DarkNet_cfg.keys(), \ "only support num_layers in {} currently" \ .format(DarkNet_cfg.keys()) self.stages = DarkNet_cfg[num_layers] self.stages = self.stages[0:5] self.num_classes = num_classes self.with_pool = True ch_in = 3 self.conv0 = ConvBNLayer( ch_in=ch_in, ch_out=32, filter_size=3, stride=1, padding=1) self.downsample0 = DownSample(ch_in=32, ch_out=32 * 2) self.darknet53_conv_block_list = [] self.downsample_list = [] ch_in = [64, 128, 256, 512, 1024] for i, stage in enumerate(self.stages): conv_block = self.add_sublayer("stage_%d" % (i), LayerWarp( int(ch_in[i]), 32 * (2**i), stage)) self.darknet53_conv_block_list.append(conv_block) for i in range(len(self.stages) - 1): downsample = self.add_sublayer( "stage_%d_downsample" % i, DownSample( ch_in=32 * (2**(i + 1)), ch_out=32 * (2**(i + 2)))) self.downsample_list.append(downsample) if self.with_pool: self.global_pool = Pool2D( pool_size=7, pool_type='avg', global_pooling=True) if self.num_classes > 0: stdv = 1.0 / math.sqrt(32 * (2**(i + 2))) self.fc_input_dim = 32 * (2**(i + 2)) self.fc = Linear( self.fc_input_dim, num_classes, act='softmax', param_attr=fluid.param_attr.ParamAttr( initializer=fluid.initializer.Uniform(-stdv, stdv))) def forward(self, inputs): out = self.conv0(inputs) out = self.downsample0(out) for i, conv_block_i in enumerate(self.darknet53_conv_block_list): out = conv_block_i(out) if i < len(self.stages) - 1: out = self.downsample_list[i](out) if self.with_pool: out = self.global_pool(out) if self.num_classes > 0: out = fluid.layers.reshape(out, shape=[-1, self.fc_input_dim]) out = self.fc(out) return out def _darknet(arch, num_layers=53, pretrained=False, **kwargs): model = DarkNet(num_layers, **kwargs) if pretrained: assert arch in model_urls, "{} model do not have a pretrained model now, you should set pretrained=False".format( arch) weight_path = get_weights_path_from_url(*(model_urls[arch])) assert weight_path.endswith('.pdparams'), \ "suffix of weight must be .pdparams" model.load(weight_path) return model def darknet53(pretrained=False, **kwargs): """DarkNet 53-layer model Args: input_channels (bool): channel number of input data, default 3. pretrained (bool): If True, returns a model pre-trained on ImageNet, default True. """ return _darknet('darknet53', 53, pretrained, **kwargs)