layout_data.models.fpn package

Submodules

layout_data.models.fpn.fpn_head module

class layout_data.models.fpn.fpn_head.Conv3x3GNReLU(in_channels, out_channels, upsample=False)

基类：sphinx.ext.autodoc.importer._MockObject

forward(x, size)

class layout_data.models.fpn.fpn_head.FPNBlock(pyramid_channels, skip_channels)

基类：sphinx.ext.autodoc.importer._MockObject

forward(x)

class layout_data.models.fpn.fpn_head.FPNDecoder(encoder_channels, pyramid_channels=256, segmentation_channels=128, final_upsampling=4, final_channels=1, dropout=0.2)

基类：sphinx.ext.autodoc.importer._MockObject

forward(x)

class layout_data.models.fpn.fpn_head.SegmentationBlock(in_channels, out_channels, n_upsamples=0)

基类：sphinx.ext.autodoc.importer._MockObject

forward(x, sizes=[])

layout_data.models.fpn.model module

class layout_data.models.fpn.model.FPNModel(hparams)

基类：sphinx.ext.autodoc.importer._MockObject

static add_model_specific_args(parser)

Parameters you define here will be available to your model through self.hparams.

configure_optimizers()

forward(x)

prepare_data()

Prepare dataset

test_dataloader()

test_epoch_end(outputs)

test_step(batch, batch_idx)

train_dataloader()

training_step(batch, batch_idx)

val_dataloader()

validation_epoch_end(outputs)

validation_step(batch, batch_idx)

layout_data.models.fpn.model_init module

layout_data.models.fpn.model_init.weights_init(m)

模型的权重初始化函数，由模型调用，如CRNN model :param m: 待初始化的模型 nn.Module :return:

layout_data.models.fpn.model_init.weights_init_without_kaiming(m)

模型的权重初始化函数，由模型调用，如CRNN model :param m: 待初始化的模型 nn.Module :return:

layout_data.models.fpn.resnet module

class layout_data.models.fpn.resnet.ResNet(block, layers)

基类：sphinx.ext.autodoc.importer._MockObject

forward(input)

layout_data.models.fpn.resnet.resnet50(pretrained=False, **kwargs)

Constructs a ResNet-50 model.

参数:pretrained (bool) -- If True, returns a model pre-trained on ImageNet

layout_data.models.fpn.resnet.resnet101(pretrained=False, **kwargs)

Constructs a ResNet-101 model.

参数:pretrained (bool) -- If True, returns a model pre-trained on ImageNet

layout_data.models.fpn.resnet.resnet152(pretrained=False, **kwargs)

Constructs a ResNet-152 model.

参数:pretrained (bool) -- If True, returns a model pre-trained on ImageNet

Module contents