o V h #@sddlmZddlmZmZddlmZddlmZddl m Z m Z m Z ddl mZdd lmZmZmZdd lmZmZmZmZmZd d lmZgd ZGdddeZGdddejZedddZGddde ZGddde Z dede!dee"defddZ#e edej$fd ej%fd!d"d#d"d"ej%d$d%eed&e"dee!d'ee"d(eed)edefd*d+Z&e ede j$fd ej%fd!d"d#d"d"ej%d$d%ee d&e"dee!d'ee"d(eed)edefd,d-Z'd"S).)partial)AnyOptional)nn)SemanticSegmentation)register_modelWeights WeightsEnum)_VOC_CATEGORIES)_ovewrite_value_paramhandle_legacy_interfaceIntermediateLayerGetter)ResNet resnet101ResNet101_Weightsresnet50ResNet50_Weights)_SimpleSegmentationModel)FCNFCN_ResNet50_WeightsFCN_ResNet101_Weights fcn_resnet50 fcn_resnet101c@seZdZdZdS)ra Implements FCN model from `"Fully Convolutional Networks for Semantic Segmentation" `_. Args: backbone (nn.Module): the network used to compute the features for the model. The backbone should return an OrderedDict[Tensor], with the key being "out" for the last feature map used, and "aux" if an auxiliary classifier is used. classifier (nn.Module): module that takes the "out" element returned from the backbone and returns a dense prediction. aux_classifier (nn.Module, optional): auxiliary classifier used during training N)__name__ __module__ __qualname____doc__r r W/var/www/vscode/kcb/lib/python3.10/site-packages/torchvision/models/segmentation/fcn.pyrsrcs*eZdZdededdffdd ZZS)FCNHead in_channelschannelsreturnNc sP|d}tj||ddddt|ttdt||dg}tj|dS)NrrF)paddingbiasg?)rConv2d BatchNorm2dReLUDropoutsuper__init__)selfr#r$inter_channelslayers __class__r r!r.%s zFCNHead.__init__)rrrintr. __classcell__r r r2r!r"$s"r")rrz These weights were trained on a subset of COCO, using only the 20 categories that are present in the Pascal VOC dataset. ) categoriesmin_size_docsc @DeZdZedeeddiedddddd id d d d ZeZdS)rzBhttps://download.pytorch.org/models/fcn_resnet50_coco-1167a1af.pth resize_sizeijzPhttps://github.com/pytorch/vision/tree/main/references/segmentation#fcn_resnet50COCO-val2017-VOC-labelsg@N@gV@miou pixel_accgmc@g?5^I`@ num_paramsrecipe_metrics_ops _file_sizeurl transformsmetaN rrrr rr _COMMON_METACOCO_WITH_VOC_LABELS_V1DEFAULTr r r r!r<& rc @r9)rzChttps://download.pytorch.org/models/fcn_resnet101_coco-7ecb50ca.pthr:r;ij<zWhttps://github.com/pytorch/vision/tree/main/references/segmentation#deeplabv3_resnet101r=gO@gV@r>gV-m@gni@rArGNrKr r r r!rQrOrbackbone num_classesauxr%cCsHddi}|r d|d<t||d}|rtd|nd}td|}t|||S)Nlayer4outrRlayer3) return_layersii)rr"r)rPrQrRrVaux_classifier classifierr r r! _fcn_resnetfs   rY pretrainedpretrained_backbone)weightsweights_backboneNT)r\progressrQaux_lossr]r\r^r_r]kwargscKt|}t|}|dur"d}td|t|jd}td|d}n|dur(d}t|gdd}t|||}|durD||j |dd |S) a\Fully-Convolutional Network model with a ResNet-50 backbone from the `Fully Convolutional Networks for Semantic Segmentation `_ paper. .. betastatus:: segmentation module Args: weights (:class:`~torchvision.models.segmentation.FCN_ResNet50_Weights`, optional): The pretrained weights to use. See :class:`~torchvision.models.segmentation.FCN_ResNet50_Weights` below for more details, and possible values. By default, no pre-trained weights are used. progress (bool, optional): If True, displays a progress bar of the download to stderr. Default is True. num_classes (int, optional): number of output classes of the model (including the background). aux_loss (bool, optional): If True, it uses an auxiliary loss. weights_backbone (:class:`~torchvision.models.ResNet50_Weights`, optional): The pretrained weights for the backbone. **kwargs: parameters passed to the ``torchvision.models.segmentation.fcn.FCN`` base class. Please refer to the `source code `_ for more details about this class. .. autoclass:: torchvision.models.segmentation.FCN_ResNet50_Weights :members: NrQr6r_TFTTr\replace_stride_with_dilationr^ check_hash) rverifyrr lenrJrrYload_state_dictget_state_dictr\r^rQr_r]r`rPmodelr r r!ru (  rcKra) aaFully-Convolutional Network model with a ResNet-101 backbone from the `Fully Convolutional Networks for Semantic Segmentation `_ paper. .. betastatus:: segmentation module Args: weights (:class:`~torchvision.models.segmentation.FCN_ResNet101_Weights`, optional): The pretrained weights to use. See :class:`~torchvision.models.segmentation.FCN_ResNet101_Weights` below for more details, and possible values. By default, no pre-trained weights are used. progress (bool, optional): If True, displays a progress bar of the download to stderr. Default is True. num_classes (int, optional): number of output classes of the model (including the background). aux_loss (bool, optional): If True, it uses an auxiliary loss. weights_backbone (:class:`~torchvision.models.ResNet101_Weights`, optional): The pretrained weights for the backbone. **kwargs: parameters passed to the ``torchvision.models.segmentation.fcn.FCN`` base class. Please refer to the `source code `_ for more details about this class. .. autoclass:: torchvision.models.segmentation.FCN_ResNet101_Weights :members: NrQr6r_Trbrcrdrf) rrhrr rirJrrYrjrkrlr r r!rrnr)( functoolsrtypingrrtorchrtransforms._presetsr_apir r r _metar _utilsr rrresnetrrrrrr__all__r Sequentialr"rLrrr4boolrYrM IMAGENET1K_V1rrr r r r!s       6