a PSic: @sddlmZddlmZmZmZddlZddlmZddlm Z ddl m Z dd l mZmZdd lmZdd lmZmZmZdd lmZmZmZdd lmZmZmZmZmZddlm Z ddl!m"Z"gdZ#Gddde Z$Gdddej%Z&Gdddej%Z'Gdddej%Z(Gdddej)Z*ee+ee,e$dddZ-edd d!Z.Gd"d#d#eZ/Gd$d%d%eZ0Gd&d'd'eZ1ee+ee,e$dd(d)Z2ed*e/j3fd+ej4fd,dd-ddej4d.ee/e,ee+ee,eeee$d/d0d1Z5ed*e0j3fd+ej4fd,dd-ddej4d.ee0e,ee+ee,eeee$d/d2d3Z6ed*e1j3fd+ej4fd,dd-ddej4d.ee1e,ee+ee,eeee$d/d4d5Z7dd6lm8Z8e8e/j3j9e0j3j9e1j3j9d7Z:dS)8)partial)AnyListOptionalN)nn) functional)SemanticSegmentation) WeightsEnumWeights)_VOC_CATEGORIES)IntermediateLayerGetterhandle_legacy_interface_ovewrite_value_param) MobileNetV3MobileNet_V3_Large_Weightsmobilenet_v3_large)ResNetresnet50 resnet101ResNet50_WeightsResNet101_Weights)_SimpleSegmentationModel)FCNHead) DeepLabV3DeepLabV3_ResNet50_WeightsDeepLabV3_ResNet101_Weights$DeepLabV3_MobileNet_V3_Large_Weightsdeeplabv3_mobilenet_v3_largedeeplabv3_resnet50deeplabv3_resnet101c@seZdZdZdS)ra Implements DeepLabV3 model from `"Rethinking Atrous Convolution for Semantic Image 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'e/var/www/html/django/DPS/env/lib/python3.9/site-packages/torchvision/models/segmentation/deeplabv3.pyrsrcs&eZdZeeddfdd ZZS) DeepLabHeadN) in_channels num_classesreturnc sFtt|gdtjddddddtdttd|ddS)N) $rrF)paddingbias)super__init__ASPPrConv2d BatchNorm2dReLU)selfr*r+ __class__r'r(r41s  zDeepLabHead.__init__r#r$r%intr4 __classcell__r'r'r:r(r)0sr)cs(eZdZeeeddfdd ZZS)ASPPConvN)r* out_channelsdilationr,cs6tj||d||ddt|tg}tj|dS)NrF)r1rAr2)rr6r7r8r3r4)r9r*r@rAmodulesr:r'r(r4<s zASPPConv.__init__r<r'r'r:r(r?;sr?cs:eZdZeeddfdd ZejejdddZZS) ASPPPoolingN)r*r@r,c s4ttdtj||dddt|tdS)NrFr2)r3r4rAdaptiveAvgPool2dr6r7r8)r9r*r@r:r'r(r4Fs zASPPPooling.__init__xr,cCs2|jdd}|D] }||}qtj||dddS)NbilinearF)sizemode align_corners)shapeF interpolate)r9rGrJmodr'r'r(forwardNs zASPPPooling.forward) r#r$r%r=r4torchTensorrQr>r'r'r:r(rCEsrCcsBeZdZd eeeeddfdd ZejejdddZZ S) r5r0N)r* atrous_ratesr@r,c stg}|ttj||dddt|tt|}|D]}|t |||qF|t ||t ||_ ttjt |j ||dddt|ttd|_dS)NrFrDg?)r3r4appendr Sequentialr6r7r8tupler?rC ModuleListconvslenDropoutproject)r9r*rTr@rBratesrater:r'r(r4Vs  $ z ASPP.__init__rFcCs6g}|jD]}|||q tj|dd}||S)Nr)dim)rYrUrRcatr\)r9rGZ_resconvresr'r'r(rQls  z ASPP.forward)r0) r#r$r%r=rr4rRrSrQr>r'r'r:r(r5Usr5)backboner+auxr,cCsHddi}|rd|d<t||d}|r.td|nd}td|}t|||S)Nlayer4outrdlayer3 return_layersii)rrr)r)rcr+rdriaux_classifier classifierr'r'r(_deeplabv3_resnetts  rl)rrz These weights were trained on a subset of COCO, using only the 20 categories that are present in the Pascal VOC dataset. ) categoriesmin_size_docsc @s@eZdZedeeddiedddddd id d ZeZd S) rzHhttps://download.pytorch.org/models/deeplabv3_resnet50_coco-cd0a2569.pth resize_sizeijzVhttps://github.com/pytorch/vision/tree/main/references/segmentation#deeplabv3_resnet50COCO-val2017-VOC-labelsgP@皙W@ZmiouZ pixel_acc num_paramsrecipe_metricsurl transformsmetaN r#r$r%r rr _COMMON_METACOCO_WITH_VOC_LABELS_V1DEFAULTr'r'r'r(rs  rc @s@eZdZedeeddiedddddd id d ZeZd S) rzIhttps://download.pytorch.org/models/deeplabv3_resnet101_coco-586e9e4e.pthrprqijzQhttps://github.com/pytorch/vision/tree/main/references/segmentation#fcn_resnet101rsgP@rtrurvrzNr~r'r'r'r(rs  rc @s@eZdZedeeddiedddddd id d ZeZd S) rzMhttps://download.pytorch.org/models/deeplabv3_mobilenet_v3_large-fc3c493d.pthrprqiPKz`https://github.com/pytorch/vision/tree/main/references/segmentation#deeplabv3_mobilenet_v3_largersgfffff&N@gV@rurvrzNr~r'r'r'r(rs  rc Cs|j}dgddt|Dt|dg}|d}||j}|d}||j}t|di}|rld|t|<t||d }|rt||nd} t||} t|| | S) NrcSs g|]\}}t|ddr|qS)_is_cnF)getattr).0ibr'r'r( z*_deeplabv3_mobilenetv3..rrfrdrh) features enumeraterZr@strrrr)r) rcr+rd stage_indicesZout_posZ out_inplanesZaux_posZ aux_inplanesrirjrkr'r'r(_deeplabv3_mobilenetv3s&      r pretrainedpretrained_backbone)weightsweights_backboneT)rprogressr+aux_lossr)rrr+rrkwargsr,cKst|}t|}|dur@d}t|t|jd}t|d}n |durLd}t|gdd}t|||}|dur||j |d|S)adConstructs a DeepLabV3 model with a ResNet-50 backbone. .. betastatus:: segmentation module Reference: `Rethinking Atrous Convolution for Semantic Image Segmentation `__. Args: weights (:class:`~torchvision.models.segmentation.DeepLabV3_ResNet50_Weights`, optional): The pretrained weights to use. See :class:`~torchvision.models.segmentation.DeepLabV3_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: unused .. autoclass:: torchvision.models.segmentation.DeepLabV3_ResNet50_Weights :members: NrmTFTTrreplace_stride_with_dilationr) rverifyrrrZr}rrlload_state_dictget_state_dictrrr+rrrrcmodelr'r'r(r!s$    r!cKst|}t|}|dur@d}t|t|jd}t|d}n |durLd}t|gdd}t|||}|dur||j |d|S)aiConstructs a DeepLabV3 model with a ResNet-101 backbone. .. betastatus:: segmentation module Reference: `Rethinking Atrous Convolution for Semantic Image Segmentation `__. Args: weights (:class:`~torchvision.models.segmentation.DeepLabV3_ResNet101_Weights`, optional): The pretrained weights to use. See :class:`~torchvision.models.segmentation.DeepLabV3_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: unused .. autoclass:: torchvision.models.segmentation.DeepLabV3_ResNet101_Weights :members: NrmTrrrr) rrrrrZr}rrlrrrr'r'r(r"s$    r"cKst|}t|}|dur@d}t|t|jd}t|d}n |durLd}t|dd}t|||}|dur~||j |d|S)akConstructs a DeepLabV3 model with a MobileNetV3-Large backbone. Reference: `Rethinking Atrous Convolution for Semantic Image Segmentation `__. Args: weights (:class:`~torchvision.models.segmentation.DeepLabV3_MobileNet_V3_Large_Weights`, optional): The pretrained weights to use. See :class:`~torchvision.models.segmentation.DeepLabV3_MobileNet_V3_Large_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.MobileNet_V3_Large_Weights`, optional): The pretrained weights for the backbone **kwargs: unused .. autoclass:: torchvision.models.segmentation.DeepLabV3_MobileNet_V3_Large_Weights :members: NrmTr)rdilatedr) rrrrrZr}rrrrrr'r'r(r Ks"     r ) _ModelURLs)Zdeeplabv3_resnet50_cocoZdeeplabv3_resnet101_cocoZ!deeplabv3_mobilenet_v3_large_coco); functoolsrtypingrrrrRrtorch.nnrrNZtransforms._presetsr _apir r _metar _utilsrrr mobilenetv3rrrresnetrrrrrrfcnr__all__rrVr)r?rCModuler5r=boolrlrrrrrr IMAGENET1K_V1r!r"r rr{ model_urlsr'r'r'r(s              332