o + iU@sUddlmZddlmZmZddlZddlmZddlmZm Z m Z ddl m Z ddl mZeredd lmZmZdd lmZmZdd lmZdd lmZed ZedZGdddeZGdddeZgZdddddddd Zded<dId d!ZdJd&d'ZGd(d)d)ej Z!Gd*d+d+ej Z"Gd,d-d-ej Z# . /dKdLd8d9Z$ .dMdNd:d;Z% .dMdNdd?Z' .dMdNd@dAZ( .dMdNdBdCZ) .dMdNdDdEZ* .dMdOdGdHZ+dS)P) annotations) TYPE_CHECKINGCallableN)nn)AdaptiveAvgPool2DLinear MaxPool2D)get_weights_path_from_url)ConvNormActivation)Literal TypedDict) NotRequiredUnpack)Tensor)Size2)shufflenet_v2_x0_25shufflenet_v2_x0_33shufflenet_v2_x0_5shufflenet_v2_x1_0shufflenet_v2_x1_5shufflenet_v2_x2_0shufflenet_v2_swish)reluswishc@s&eZdZUded<ded<ded<dS)_ShuffleNetOptionsz#NotRequired[_ActivationType | None]actNotRequired[bool] with_poolNotRequired[int] num_classesN__name__ __module__ __qualname____annotations__r&r&m/home/app/PaddleOCR-VL-test/.venv_paddleocr/lib/python3.10/site-packages/paddle/vision/models/shufflenetv2.pyr.s  rc@seZdZUded<ded<dS)_ShuffleNetSwishOptionsrrrr Nr!r&r&r&r'r(3s  r()zEhttps://paddle-hapi.bj.bcebos.com/models/shufflenet_v2_x0_25.pdparamsZ 1e509b4c140eeb096bb16e214796d03b)zEhttps://paddle-hapi.bj.bcebos.com/models/shufflenet_v2_x0_33.pdparamsZ 3d7b3ab0eaa5c0927ff1026d31b729bd)zDhttps://paddle-hapi.bj.bcebos.com/models/shufflenet_v2_x0_5.pdparamsZ 5e5cee182a7793c4e4c73949b1a71bd4)zDhttps://paddle-hapi.bj.bcebos.com/models/shufflenet_v2_x1_0.pdparamsZ 122d42478b9e81eb49f8a9ede327b1a4)zDhttps://paddle-hapi.bj.bcebos.com/models/shufflenet_v2_x1_5.pdparamsZ faced5827380d73531d0ee027c67826d)zDhttps://paddle-hapi.bj.bcebos.com/models/shufflenet_v2_x2_0.pdparamsZ cd3dddcd8305e7bcd8ad14d1c69a5784)zEhttps://paddle-hapi.bj.bcebos.com/models/shufflenet_v2_swish.pdparamsZ adde0aa3b023e5b0c94a68be1c394b84zdict[str, tuple[str, str]] model_urlsr_ActivationType | Nonereturnnn.Layer | NonecCs6|dkrtjS|dkrtjS|durdStd|)Nrrz*The activation function is not supported: )rZSwishReLU RuntimeError)rr&r&r'create_activation_layerZsr/xrgroupsintcCsb|jdd\}}}}||}tj||||||gd}tj|gdd}tj|||||gd}|S)Nr)shape)rr r3)perm)r4paddleZreshapeZ transpose)r0r1Z batch_sizeZ num_channelsheightwidthZchannels_per_groupr&r&r'channel_shuffleesr;c.eZdZejfdfd d ZdddZZS)InvertedResidual in_channelsr2 out_channelsstrideractivation_layerCallable[..., nn.Layer]r+Nonec srtt|d|ddddd|d|_t|d|dd|d|ddd|_t|d|ddddd|d|_dS)Nr r5rr>r? kernel_sizer@paddingr1rAr6)super__init__r _conv_pw_conv_dw _conv_linearselfr>r?r@rA __class__r&r'rHws8    zInvertedResidual.__init__inputsrcCsftj||jdd|jddgdd\}}||}||}||}tj||gdd}t|dS)Nr5r )Znum_or_sectionsaxisrQ)r8splitr4rIrJrKconcatr;rMrPx1Zx2outr&r&r'forwards    zInvertedResidual.forward r>r2r?r2r@rrArBr+rCrPrr+rr"r#r$rr-rHrX __classcell__r&r&rNr'r=vs$r=cr<)InvertedResidualDSr>r2r?r@rrArBr+rCc stt||d|d|dd|_t||ddddd|d|_t||ddddd|d|_t|d|dd|d|ddd|_t|d|ddddd|d|_dS)Nr6r5rDr r)rGrHr _conv_dw_1_conv_linear_1 _conv_pw_2 _conv_dw_2_conv_linear_2rLrNr&r'rHs\      zInvertedResidualDS.__init__rPrcCsN||}||}||}||}||}tj||gdd}t|dS)Nr5rRr )r^r_r`rarbr8rTr;rUr&r&r'rXs      zInvertedResidualDS.forwardrYrZr[r&r&rNr'r]s9r]csPeZdZUdZded<ded<ded<  ddfdd ZdddZZS) ShuffleNetV2aShuffleNetV2 model from `"ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Design" `_. Args: scale (float, optional): Scale of output channels. Default: True. act (str, optional): Activation function of neural network. Default: "relu". 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. Returns: :ref:`api_paddle_nn_Layer`. An instance of ShuffleNetV2 model. Examples: .. code-block:: python >>> import paddle >>> from paddle.vision.models import ShuffleNetV2 >>> shufflenet_v2_swish = ShuffleNetV2(scale=1.0, act="swish") >>> x = paddle.rand([1, 3, 224, 224]) >>> out = shufflenet_v2_swish(x) >>> print(out.shape) [1, 1000] floatscaler2r boolr?rTrr*r+rCc st||_||_||_gd}t|}|dkrgd}n7|dkr(gd}n.|dkr1gd}n%|dkr:gd }n|d krCgd }n|d krLgd }n tdt|dtd|dddd|d|_ t dddd|_ g|_ t |D]X\}} t| D]O} | dkr|jt||d||dd|dt|ddt| dd} n!|jt||d||dd|dt|ddt| dd} |j | qzqrt|d|dddd|d|_|rtd|_|dkr|d|_t|d||_dSdS)N)r3r3?)rl0`Q?)rkrl @ro?)rkrlrmrnrg)rkrltirv?)rkrli`irv@)rkrliiizThis scale size:[z] is not implemented!r6r5r )r>r?rEr@rFrA)rEr@rFr)r>r?r@rA_)Zsublayernamerk)rGrHrer rr/NotImplementedErrorstrr _conv1r _max_pool _block_list enumeraterangeZ add_sublayerr]r=append _last_convr _pool2d_avgZ_out_cr_fc) rMrerr rZ stage_repeatsrAZstage_out_channelsZstage_idZ num_repeatiblockrNr&r'rH s                zShuffleNetV2.__init__rPrcCsj||}||}|jD]}||}q ||}|jr!||}|jdkr3tj|ddd}| |}|S)Nrr5rk)Z start_axisZ stop_axis) rrrrrrr r8flattenr)rMrPr0invr&r&r'rX`s        zShuffleNetV2.forward)rgrrhT) rerdrr*r r2rrfr+rCrZ)r"r#r$__doc__r%rHrXr\r&r&rNr'rcs TrcFrgarch_ShuffleNetArch pretrainedrfrerdkwargsUnpack[_ShuffleNetOptions]cKs^tdd|i|}|r-|tvsJ|dtt|dt|d}t|}|||S)NrezJ model do not have a pretrained model now, you should set pretrained=Falserr5r&)rcr)r r8loadZset_dict)rrrermodelZ weight_pathparamr&r&r'_shufflenet_v2ps   rcKt dd|d|S)aUShuffleNetV2 with 0.25x output channels, as described in `"ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Design" `_. 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:`ShuffleNetV2 `. Returns: :ref:`api_paddle_nn_Layer`. An instance of ShuffleNetV2 with 0.25x output channels. Examples: .. code-block:: python >>> import paddle >>> from paddle.vision.models import shufflenet_v2_x0_25 >>> # build model >>> model = shufflenet_v2_x0_25() >>> # build model and load imagenet pretrained weight >>> # model = shufflenet_v2_x0_25(pretrained=True) >>> x = paddle.rand([1, 3, 224, 224]) >>> out = model(x) >>> print(out.shape) [1, 1000] rrjrerN)rrrrr&r&r'r rcKr)aUShuffleNetV2 with 0.33x output channels, as described in `"ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Design" `_. 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:`ShuffleNetV2 `. Returns: :ref:`api_paddle_nn_Layer`. An instance of ShuffleNetV2 with 0.33x output channels. Examples: .. code-block:: python >>> import paddle >>> from paddle.vision.models import shufflenet_v2_x0_33 >>> # build model >>> model = shufflenet_v2_x0_33() >>> # build model and load imagenet pretrained weight >>> # model = shufflenet_v2_x0_33(pretrained=True) >>> x = paddle.rand([1, 3, 224, 224]) >>> out = model(x) >>> print(out.shape) [1, 1000] rrprN)rrrr&r&r'rrrcKr)aPShuffleNetV2 with 0.5x output channels, as described in `"ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Design" `_. 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:`ShuffleNetV2 `. Returns: :ref:`api_paddle_nn_Layer`. An instance of ShuffleNetV2 with 0.5x output channels. Examples: .. code-block:: python >>> import paddle >>> from paddle.vision.models import shufflenet_v2_x0_5 >>> # build model >>> model = shufflenet_v2_x0_5() >>> # build model and load imagenet pretrained weight >>> # model = shufflenet_v2_x0_5(pretrained=True) >>> x = paddle.rand([1, 3, 224, 224]) >>> out = model(x) >>> print(out.shape) [1, 1000] rrtrN)rrrr&r&r'rrrcKr)aPShuffleNetV2 with 1.0x output channels, as described in `"ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Design" `_. 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:`ShuffleNetV2 `. Returns: :ref:`api_paddle_nn_Layer`. An instance of ShuffleNetV2 with 1.0x output channels. Examples: .. code-block:: python >>> import paddle >>> from paddle.vision.models import shufflenet_v2_x1_0 >>> # build model >>> model = shufflenet_v2_x1_0() >>> # build model and load imagenet pretrained weight >>> # model = shufflenet_v2_x1_0(pretrained=True) >>> x = paddle.rand([1, 3, 224, 224]) >>> out = model(x) >>> print(out.shape) [1, 1000] rrgrN)rrrr&r&r'rrrcKr)aPShuffleNetV2 with 1.5x output channels, as described in `"ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Design" `_. 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:`ShuffleNetV2 `. Returns: :ref:`api_paddle_nn_Layer`. An instance of ShuffleNetV2 with 1.5x output channels. Examples: .. code-block:: python >>> import paddle >>> from paddle.vision.models import shufflenet_v2_x1_5 >>> # build model >>> model = shufflenet_v2_x1_5() >>> # build model and load imagenet pretrained weight >>> # model = shufflenet_v2_x1_5(pretrained=True) >>> x = paddle.rand([1, 3, 224, 224]) >>> out = model(x) >>> print(out.shape) [1, 1000] rryrN)rrrr&r&r'rrrcKr)aPShuffleNetV2 with 2.0x output channels, as described in `"ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Design" `_. 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:`ShuffleNetV2 `. Returns: :ref:`api_paddle_nn_Layer`. An instance of ShuffleNetV2 with 2.0x output channels. Examples: .. code-block:: python >>> import paddle >>> from paddle.vision.models import shufflenet_v2_x2_0 >>> # build model >>> model = shufflenet_v2_x2_0() >>> # build model and load imagenet pretrained weight >>> # model = shufflenet_v2_x2_0(pretrained=True) >>> x = paddle.rand([1, 3, 224, 224]) >>> out = model(x) >>> print(out.shape) [1, 1000] rr{rN)rrrr&r&r'r=rrUnpack[_ShuffleNetSwishOptions]cKst ddd|d|S)a]ShuffleNetV2 with swish activation function, as described in `"ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Design" `_. 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:`ShuffleNetV2 `. Returns: :ref:`api_paddle_nn_Layer`. An instance of ShuffleNetV2 with swish activation function. Examples: .. code-block:: python >>> import paddle >>> from paddle.vision.models import shufflenet_v2_swish >>> # build model >>> model = shufflenet_v2_swish() >>> # build model and load imagenet pretrained weight >>> # model = shufflenet_v2_swish(pretrained=True) >>> x = paddle.rand([1, 3, 224, 224]) >>> out = model(x) >>> print(out.shape) [1, 1000] rrgr)rerrN)rrrr&r&r'rbs r)rr*r+r,)r0rr1r2r+r)Frg) rrrrfrerdrrr+rc)F)rrfrrr+rc)rrfrrr+rc), __future__rtypingrrr8rZ paddle.nnrrrZpaddle.utils.downloadr opsr r r Ztyping_extensionsrrrZpaddle._typingrrZ_ActivationTyperr(__all__r)r%r/r;ZLayerr=r]rcrrrrrrrrr&r&r&r'sd       2E  & & & & & &