def __init__(self, inplanes, planes, stride=1, downsample=None):
super(IRBlock, self).__init__()
self.conv1 = conv3x3(inplanes, planes, stride=stride)
self.bn1 = bn_with_initialize(planes)
self.relu1 = P.ReLU()
self.conv2 = conv3x3(planes, planes, stride=1)
self.bn2 = bn_with_initialize(planes)
if downsample is None:
self.downsample = Cut()
else:
self.downsample = downsample
self.add = TensorAdd()
self.cast = P.Cast()
self.relu2 = P.ReLU()
def __init__(self, inplanes, planes, expansion, stride):
super(DownSample, self).__init__()
self.conv1 = conv1x1(inplanes,
planes * expansion,
stride=stride,
pad_mode="valid")
self.bn1 = bn_with_initialize(planes * expansion)
def __init__(self, block, layers, flat_dim, fc_dim, attri_num_list):
super(AttriResNet, self).__init__()
# resnet18
self.inplanes = 32
self.conv1 = conv3x3(3, self.inplanes, stride=1)
self.bn1 = bn_with_initialize(self.inplanes)
self.relu = P.ReLU()
self.layer1 = MakeLayer(block,
inplanes=32,
planes=64,
blocks=layers[0],
stride=2)
self.layer2 = MakeLayer(block,
inplanes=64,
planes=128,
blocks=layers[1],
stride=2)
self.layer3 = MakeLayer(block,
inplanes=128,
planes=256,
blocks=layers[2],
stride=2)
self.layer4 = MakeLayer(block,
inplanes=256,
planes=512,
blocks=layers[3],
stride=2)
# avg global pooling
self.mean = P.ReduceMean(keep_dims=True)
self.shape = P.Shape()
self.reshape = P.Reshape()
self.head = get_attri_head(flat_dim, fc_dim, attri_num_list)