o #je@sddlZddlmZddlmZgZdddddd d d d d dddd ZGdddejZGdddejZGdddejZ ddZ d5ddZ d5ddZ d5dd Z d5d!d"Zd5d#d$Zd5d%d&Zd5d'd(Zd5d)d*Zd5d+d,Zd5d-d.Zd5d/d0Zd5d1d2Zd5d3d4ZdS)6N)nn)get_weights_path_from_url)z:https://paddle-hapi.bj.bcebos.com/models/resnet18.pdparamsZ cf548f46534aa3560945be4b95cd11c4)z:https://paddle-hapi.bj.bcebos.com/models/resnet34.pdparamsZ 8d2275cf8706028345f78ac0e1d31969)z:https://paddle-hapi.bj.bcebos.com/models/resnet50.pdparamsZ ca6f485ee1ab0492d38f323885b0ad80)z;https://paddle-hapi.bj.bcebos.com/models/resnet101.pdparamsZ 02f35f034ca3858e1e54d4036443c92d)z;https://paddle-hapi.bj.bcebos.com/models/resnet152.pdparamsZ 7ad16a2f1e7333859ff986138630fd7a)zAhttps://paddle-hapi.bj.bcebos.com/models/resnext50_32x4d.pdparamsZ dc47483169be7d6f018fcbb7baf8775d)zAhttps://paddle-hapi.bj.bcebos.com/models/resnext50_64x4d.pdparamsZ 063d4b483e12b06388529450ad7576db)zBhttps://paddle-hapi.bj.bcebos.com/models/resnext101_32x4d.pdparamsZ 967b090039f9de2c8d06fe994fb9095f)zBhttps://paddle-hapi.bj.bcebos.com/models/resnext101_64x4d.pdparamsZ 98e04e7ca616a066699230d769d03008)zBhttps://paddle-hapi.bj.bcebos.com/models/resnext152_32x4d.pdparamsZ 18ff0beee21f2efc99c4b31786107121)zBhttps://paddle-hapi.bj.bcebos.com/models/resnext152_64x4d.pdparamsZ 77c4af00ca42c405fa7f841841959379)zAhttps://paddle-hapi.bj.bcebos.com/models/wide_resnet50_2.pdparamsZ 0282f804d73debdab289bd9fea3fa6dc)zBhttps://paddle-hapi.bj.bcebos.com/models/wide_resnet101_2.pdparamsZ d4360a2d23657f059216f5d5a1a9ac93) resnet18resnet34resnet50 resnet101 resnet152resnext50_32x4dresnext50_64x4dresnext101_32x4dresnext101_64x4dresnext152_32x4dresnext152_64x4dwide_resnet50_2wide_resnet101_2cs6eZdZdZ      dfdd ZddZZS) BasicBlockN@c st|dur tj}|dkrtdtj||dd|dd|_|||_t|_ tj||dddd|_ |||_ ||_ ||_ dS)Nrz(Dilation > 1 not supported in BasicBlockF)paddingstride bias_attr)rr)super__init__r BatchNorm2DNotImplementedErrorConv2Dconv1bn1ReLUreluconv2bn2 downsampler) selfinplanesplanesrr#groups base_widthdilation norm_layer __class__\/var/www/html/Deteccion_Ine/venv/lib/python3.10/site-packages/paddle/vision/models/resnet.pyrPs      zBasicBlock.__init__cCs`|}||}||}||}||}||}|jdur%||}||7}||}|SN)rrr r!r"r#r$xidentityoutr-r-r.forwardns        zBasicBlock.forwardrNrrrN__name__ __module__ __qualname__ expansionrr4 __classcell__r-r-r+r.rMsrcs6eZdZdZ      d fdd ZddZZS) BottleneckBlockrNrc st|dur tj}t||d|} tj|| ddd|_|| |_tj| | d||||dd|_|| |_ tj| ||j ddd|_ |||j |_ t |_||_||_dS)NgP@rF)rr)rrr'r)r)rrrrintrrrr!r"r:conv3bn3rr r#r) r$r%r&rr#r'r(r)r*widthr+r-r.rs0     zBottleneckBlock.__init__cCs~|}||}||}||}||}||}||}||}||}|jdur4||}||7}||}|Sr/)rrr r!r"r?r@r#r0r-r-r.r4s           zBottleneckBlock.forwardr5r6r-r-r+r.r<s'r<cs>eZdZdZ     dfdd Zdd d Zd d ZZS)ResNeta.ResNet model from `"Deep Residual Learning for Image Recognition" `_. Args: Block (BasicBlock|BottleneckBlock): Block module of model. depth (int, optional): Layers of ResNet, Default: 50. width (int, optional): Base width per convolution group for each convolution block, Default: 64. num_classes (int, optional): Output dim of last fc layer. If num_classes <= 0, last fc layer will not be defined. Default: 1000. with_pool (bool, optional): Use pool before the last fc layer or not. Default: True. groups (int, optional): Number of groups for each convolution block, Default: 1. Returns: :ref:`api_paddle_nn_Layer`. An instance of ResNet model. Examples: .. code-block:: python >>> import paddle >>> from paddle.vision.models import ResNet >>> from paddle.vision.models.resnet import BottleneckBlock, BasicBlock >>> # build ResNet with 18 layers >>> resnet18 = ResNet(BasicBlock, 18) >>> # build ResNet with 50 layers >>> resnet50 = ResNet(BottleneckBlock, 50) >>> # build Wide ResNet model >>> wide_resnet50_2 = ResNet(BottleneckBlock, 50, width=64*2) >>> # build ResNeXt model >>> resnext50_32x4d = ResNet(BottleneckBlock, 50, width=4, groups=32) >>> x = paddle.rand([1, 3, 224, 224]) >>> out = resnet18(x) >>> print(out.shape) [1, 1000] 2rTrc s6tgdgdgdgdgdd}||}||_||_||_||_tj|_d|_ d|_ tj d|j d d dd d |_ ||j |_ t|_tjdd dd |_||d|d|_|j|d|dd d|_|j|d|d d d|_|j|d|dd d|_|rtd|_|dkrtd|j||_dSdS)N)rErErE)rr=r)rr=r)r$r)"rCerrrrEF) kernel_sizerrr)rOrrr)ri)rr)rrr'r( num_classes with_poolrr _norm_layerr%r)rrrrr Z MaxPool2Dmaxpool _make_layerlayer1layer2layer3layer4ZAdaptiveAvgPool2DavgpoolZLinearr:fc) r$blockdepthrArRrSr'Z layer_cfglayersr+r-r.rsF   zResNet.__init__Fc Cs|j}d}|j}|r|j|9_d}|dks|j||jkr6ttj|j||jd|dd|||j}g} | ||j||||j|j ||||j|_t d|D]} | ||j||j|j |dqTtj| S)NrF)rr)r'r(r*) rTr)r%r:rZ Sequentialrappendr'r(range) r$r]r&blocksrZdilater*r#Zprevious_dilationr__r-r-r.rVsT    zResNet._make_layercCs||}||}||}||}||}||}||}||}|jr0| |}|j dkr@t |d}| |}|S)Nrr)rrr rUrWrXrYrZrSr[rRpaddleflattenr\)r$r1r-r-r.r4Js            zResNet.forward)rCrrDTr)rF)r7r8r9__doc__rrVr4r;r-r-r+r.rBs, 0.rBcKs^t||fi|}|r-|tvsJ|dtt|dt|d}t|}|||S)NzJ model do not have a pretrained model now, you should set pretrained=Falserr)rB model_urlsrrdloadZset_dict)archZBlockr^ pretrainedkwargsmodelZ weight_pathparamr-r-r._resnet^s   rnFcKtdtd|fi|S)aResNet 18-layer model from `"Deep Residual Learning for Image Recognition" `_. Args: pretrained (bool, optional): Whether to load pre-trained weights. If True, returns a model pre-trained on ImageNet. Default: False. **kwargs (optional): Additional keyword arguments. For details, please refer to :ref:`ResNet `. Returns: :ref:`api_paddle_nn_Layer`. An instance of ResNet 18-layer model. Examples: .. code-block:: python >>> import paddle >>> from paddle.vision.models import resnet18 >>> # build model >>> model = resnet18() >>> # build model and load imagenet pretrained weight >>> # model = resnet18(pretrained=True) >>> x = paddle.rand([1, 3, 224, 224]) >>> out = model(x) >>> print(out.shape) [1, 1000] rrJrnrrjrkr-r-r.rnrcKro)aResNet 34-layer model from `"Deep Residual Learning for Image Recognition" `_. Args: pretrained (bool, optional): Whether to load pre-trained weights. If True, returns a model pre-trained on ImageNet. Default: False. **kwargs (optional): Additional keyword arguments. For details, please refer to :ref:`ResNet `. Returns: :ref:`api_paddle_nn_Layer`. An instance of ResNet 34-layer model. Examples: .. code-block:: python >>> import paddle >>> from paddle.vision.models import resnet34 >>> # build model >>> model = resnet34() >>> # build model and load imagenet pretrained weight >>> # model = resnet34(pretrained=True) >>> x = paddle.rand([1, 3, 224, 224]) >>> out = model(x) >>> print(out.shape) [1, 1000] rrKrprqr-r-r.rrrrcKro)aResNet 50-layer model from `"Deep Residual Learning for Image Recognition" `_. Args: pretrained (bool, optional): Whether to load pre-trained weights. If True, returns a model pre-trained on ImageNet. Default: False. **kwargs (optional): Additional keyword arguments. For details, please refer to :ref:`ResNet `. Returns: :ref:`api_paddle_nn_Layer`. An instance of ResNet 50-layer model. Examples: .. code-block:: python >>> import paddle >>> from paddle.vision.models import resnet50 >>> # build model >>> model = resnet50() >>> # build model and load imagenet pretrained weight >>> # model = resnet50(pretrained=True) >>> x = paddle.rand([1, 3, 224, 224]) >>> out = model(x) >>> print(out.shape) [1, 1000] rrCrnr<rqr-r-r.rrrrcKro)aResNet 101-layer model from `"Deep Residual Learning for Image Recognition" `_. Args: pretrained (bool, optional): Whether to load pre-trained weights. If True, returns a model pre-trained on ImageNet. Default: False. **kwargs (optional): Additional keyword arguments. For details, please refer to :ref:`ResNet `. Returns: :ref:`api_paddle_nn_Layer`. An instance of ResNet 101-layer. Examples: .. code-block:: python >>> import paddle >>> from paddle.vision.models import resnet101 >>> # build model >>> model = resnet101() >>> # build model and load imagenet pretrained weight >>> # model = resnet101(pretrained=True) >>> x = paddle.rand([1, 3, 224, 224]) >>> out = model(x) >>> print(out.shape) [1, 1000] rrLrsrqr-r-r.rrrrcKro)aResNet 152-layer model from `"Deep Residual Learning for Image Recognition" `_. Args: pretrained (bool, optional): Whether to load pre-trained weights. If True, returns a model pre-trained on ImageNet. Default: False. **kwargs (optional): Additional keyword arguments. For details, please refer to :ref:`ResNet `. Returns: :ref:`api_paddle_nn_Layer`. An instance of ResNet 152-layer model. Examples: .. code-block:: python >>> import paddle >>> from paddle.vision.models import resnet152 >>> # build model >>> model = resnet152() >>> # build model and load imagenet pretrained weight >>> # model = resnet152(pretrained=True) >>> x = paddle.rand([1, 3, 224, 224]) >>> out = model(x) >>> print(out.shape) [1, 1000] rrMrsrqr-r-r.rrrrcK&d|d<d|d<tdtd|fi|S)aResNeXt-50 32x4d model from `"Aggregated Residual Transformations for Deep Neural Networks" `_. Args: pretrained (bool, optional): Whether to load pre-trained weights. If True, returns a model pre-trained on ImageNet. Default: False. **kwargs (optional): Additional keyword arguments. For details, please refer to :ref:`ResNet `. Returns: :ref:`api_paddle_nn_Layer`. An instance of ResNeXt-50 32x4d model. Examples: .. code-block:: python >>> import paddle >>> from paddle.vision.models import resnext50_32x4d >>> # build model >>> model = resnext50_32x4d() >>> # build model and load imagenet pretrained weight >>> # model = resnext50_32x4d(pretrained=True) >>> x = paddle.rand([1, 3, 224, 224]) >>> out = model(x) >>> print(out.shape) [1, 1000] r'r=rAr rCrsrqr-r-r.r r cKrt)aResNeXt-50 64x4d model from `"Aggregated Residual Transformations for Deep Neural Networks" `_. Args: pretrained (bool, optional): Whether to load pre-trained weights. If True, returns a model pre-trained on ImageNet. Default: False. **kwargs (optional): Additional keyword arguments. For details, please refer to :ref:`ResNet `. Returns: :ref:`api_paddle_nn_Layer`. An instance of ResNeXt-50 64x4d model. Examples: .. code-block:: python >>> import paddle >>> from paddle.vision.models import resnext50_64x4d >>> # build model >>> model = resnext50_64x4d() >>> # build model and load imagenet pretrained weight >>> # model = resnext50_64x4d(pretrained=True) >>> x = paddle.rand([1, 3, 224, 224]) >>> out = model(x) >>> print(out.shape) [1, 1000] rr'r=rAr rCrsrqr-r-r.r 6rvr cKrt)aResNeXt-101 32x4d model from `"Aggregated Residual Transformations for Deep Neural Networks" `_. Args: pretrained (bool, optional): Whether to load pre-trained weights. If True, returns a model pre-trained on ImageNet. Default: False. **kwargs (optional): Additional keyword arguments. For details, please refer to :ref:`ResNet `. Returns: :ref:`api_paddle_nn_Layer`. An instance of ResNeXt-101 32x4d model. Examples: .. code-block:: python >>> import paddle >>> from paddle.vision.models import resnext101_32x4d >>> # build model >>> model = resnext101_32x4d() >>> # build model and load imagenet pretrained weight >>> # model = resnext101_32x4d(pretrained=True) >>> x = paddle.rand([1, 3, 224, 224]) >>> out = model(x) >>> print(out.shape) [1, 1000] rur'r=rAr rLrsrqr-r-r.r Yr cKrt)aResNeXt-101 64x4d model from `"Aggregated Residual Transformations for Deep Neural Networks" `_. Args: pretrained (bool, optional): Whether to load pre-trained weights. If True, returns a model pre-trained on ImageNet. Default: False. **kwargs (optional): Additional keyword arguments. For details, please refer to :ref:`ResNet `. Returns: :ref:`api_paddle_nn_Layer`. An instance of ResNeXt-101 64x4d model. Examples: .. code-block:: python >>> import paddle >>> from paddle.vision.models import resnext101_64x4d >>> # build model >>> model = resnext101_64x4d() >>> # build model and load imagenet pretrained weight >>> # model = resnext101_64x4d(pretrained=True) >>> x = paddle.rand([1, 3, 224, 224]) >>> out = model(x) >>> print(out.shape) [1, 1000] rr'r=rAr rLrsrqr-r-r.r ~rwr cKrt)aResNeXt-152 32x4d model from `"Aggregated Residual Transformations for Deep Neural Networks" `_. Args: pretrained (bool, optional): Whether to load pre-trained weights. If True, returns a model pre-trained on ImageNet. Default: False. **kwargs (optional): Additional keyword arguments. For details, please refer to :ref:`ResNet `. Returns: :ref:`api_paddle_nn_Layer`. An instance of ResNeXt-152 32x4d model. Examples: .. code-block:: python >>> import paddle >>> from paddle.vision.models import resnext152_32x4d >>> # build model >>> model = resnext152_32x4d() >>> # build model and load imagenet pretrained weight >>> # model = resnext152_32x4d(pretrained=True) >>> x = paddle.rand([1, 3, 224, 224]) >>> out = model(x) >>> print(out.shape) [1, 1000] rur'r=rAr rMrsrqr-r-r.r rwr cKrt)aResNeXt-152 64x4d model from `"Aggregated Residual Transformations for Deep Neural Networks" `_. Args: pretrained (bool, optional): Whether to load pre-trained weights. If True, returns a model pre-trained on ImageNet. Default: False. **kwargs (optional): Additional keyword arguments. For details, please refer to :ref:`ResNet `. Returns: :ref:`api_paddle_nn_Layer`. An instance of ResNeXt-152 64x4d model. Examples: .. code-block:: python >>> import paddle >>> from paddle.vision.models import resnext152_64x4d >>> # build model >>> model = resnext152_64x4d() >>> # build model and load imagenet pretrained weight >>> # model = resnext152_64x4d(pretrained=True) >>> x = paddle.rand([1, 3, 224, 224]) >>> out = model(x) >>> print(out.shape) [1, 1000] rr'r=rArrMrsrqr-r-r.rrwrcKd|d<tdtd|fi|S)aWide ResNet-50-2 model from `"Wide Residual Networks" `_. Args: pretrained (bool, optional): Whether to load pre-trained weights. If True, returns a model pre-trained on ImageNet. Default: False. **kwargs (optional): Additional keyword arguments. For details, please refer to :ref:`ResNet `. Returns: :ref:`api_paddle_nn_Layer`. An instance of Wide ResNet-50-2 model. Examples: .. code-block:: python >>> import paddle >>> from paddle.vision.models import wide_resnet50_2 >>> # build model >>> model = wide_resnet50_2() >>> # build model and load imagenet pretrained weight >>> # model = wide_resnet50_2(pretrained=True) >>> x = paddle.rand([1, 3, 224, 224]) >>> out = model(x) >>> print(out.shape) [1, 1000] rPrArrCrsrqr-r-r.rsrcKrx)aWide ResNet-101-2 model from `"Wide Residual Networks" `_. Args: pretrained (bool, optional): Whether to load pre-trained weights. If True, returns a model pre-trained on ImageNet. Default: False. **kwargs (optional): Additional keyword arguments. For details, please refer to :ref:`ResNet `. Returns: :ref:`api_paddle_nn_Layer`. An instance of Wide ResNet-101-2 model. Examples: .. code-block:: python >>> import paddle >>> from paddle.vision.models import wide_resnet101_2 >>> # build model >>> model = wide_resnet101_2() >>> # build model and load imagenet pretrained weight >>> # model = wide_resnet101_2(pretrained=True) >>> x = paddle.rand([1, 3, 224, 224]) >>> out = model(x) >>> print(out.shape) [1, 1000] rPrArrLrsrqr-r-r.rs r)F)rdrZpaddle.utils.downloadr__all__rgZLayerrr<rBrnrrrrrr r r r r rrrr-r-r-r.sH  84A  ! ! ! ! ! # # % % % %"