:j5ddlZddlmZddlmZmZmZddlZddlmZddl m Z m Z ejjjZGddejjZGd d ejj"ZGd d eZGd deZGddejjZGddejj"ZGddejjZy)N)Sequence)CallableOptionalUnion)Tensor)_log_api_usage_once _make_ntupleceZdZdZ ddedeffd Zdededede d e ed e ed e effd Z d e de fdZ defdZxZS)FrozenBatchNorm2da! BatchNorm2d where the batch statistics and the affine parameters are fixed Args: num_features (int): Number of features ``C`` from an expected input of size ``(N, C, H, W)`` eps (float): a value added to the denominator for numerical stability. Default: 1e-5 num_featuresepscnt|t|||_|j dt j ||j dt j||j dt j||j dt j |y)Nweightbias running_mean running_var)super__init__r rregister_buffertorchoneszeros)selfr r __class__s i/home/mcse/projects/srt_converter/srt-converter-venv/lib/python3.12/site-packages/torchvision/ops/misc.pyrzFrozenBatchNorm2d.__init__s D! Xuzz,'?@ VU[[%>? ^U[[-FG ]EJJ|,DE state_dictprefixlocal_metadatastrict missing_keysunexpected_keys error_msgsc H|dz}||vr||=t ||||||||y)Nnum_batches_tracked)r_load_from_state_dict) rrrr r!r"r#r$num_batches_tracked_keyrs rr'z'FrozenBatchNorm2d._load_from_state_dict$s?#)+@"@ "j 023 %  oWa rxreturncR|jjdddd}|jjdddd}|jjdddd}|jjdddd}|||j zj z}|||zz }||z|zS)N)rreshaperrrrrsqrt)rr)wbrvrmscalers rforwardzFrozenBatchNorm2d.forward6s KK  2q! , II  aQ *    % %aQ 2    & &q"a 3R$((]))++2:~5y4rc|jjd|jjdd|jdS)N(rz, eps=))r__name__rshaper)rs r__repr__zFrozenBatchNorm2d.__repr__As;..))*!DKK,=,=a,@+AzQRSSr)gh㈵>)r9 __module__ __qualname____doc__intfloatrdictstrboollistr'rr5r; __classcell__rs@rr r s F F F       3i  c I $    F  T#Trr ceZdZddddejj ejj dddejjf dedede ee edffd e ee edffd e e ee edfe fd ed e e dejjfd e e dejjfde ee edffde ede ede dejjfddffd ZxZS)ConvNormActivationr,NT in_channels out_channels kernel_size.stridepaddinggroups norm_layeractivation_layerdilationinplacer conv_layerr*c N |ttrt tr dz dz z}n\ttr tn t } t | t |  t fdt | D}| |du} | |||| || g}||j|||| ind| i}|j|di|t|$|t|||_ |jtk(rtjdyy)Nr,rc3@K|]}|dz dz|zyw)r,rN).0irRrLs r z.ConvNormActivation.__init__..]s(bAQ!!3 9HQK Gbs)rRrOrrSzhDon't use ConvNormActivation directly, please use Conv2dNormActivation and Conv3dNormActivation instead.rW) isinstancer?rlenr tuplerangeappendrrr rKrrHwarningswarn)rrJrKrLrMrNrOrPrQrRrSrrT _conv_dimlayersparamsrs ` ` rrzConvNormActivation.__init__Fs5 ?+s+ 8S0I&?q08;0:;0QC ,WZ[cWd *; B ')<bQVW`Qabb <%D !     ! MM*\2 3  '"?RG0DF MM*4V4 5 &!D!( >>/ / MMz  0r)r9r<r=rnn BatchNorm2dReLUConv2dr?rr]rrBrModulerCrrErFs@rrHrHEs^ 45./>B?Dxx?S?SEJXX]]01"&#5:XX__5553c3h/0 5 c5c?*+ 5 %U38_c 9:; 55Xc588??&:;<5#8C,@#AB5U38_,-5$5tn5S%((//125 55rrHceZdZdZddddej j ej jdddf dedede ee eeffd e ee eeffd e e ee eefe fd ed e e d ej jfde e d ej jfde ee eeffde ede eddffd ZxZS)Conv2dNormActivationa Configurable block used for Convolution2d-Normalization-Activation blocks. Args: in_channels (int): Number of channels in the input image out_channels (int): Number of channels produced by the Convolution-Normalization-Activation block kernel_size: (int, optional): Size of the convolving kernel. Default: 3 stride (int, optional): Stride of the convolution. Default: 1 padding (int, tuple or str, optional): Padding added to all four sides of the input. Default: None, in which case it will be calculated as ``padding = (kernel_size - 1) // 2 * dilation`` groups (int, optional): Number of blocked connections from input channels to output channels. Default: 1 norm_layer (Callable[..., torch.nn.Module], optional): Norm layer that will be stacked on top of the convolution layer. If ``None`` this layer won't be used. Default: ``torch.nn.BatchNorm2d`` activation_layer (Callable[..., torch.nn.Module], optional): Activation function which will be stacked on top of the normalization layer (if not None), otherwise on top of the conv layer. If ``None`` this layer won't be used. Default: ``torch.nn.ReLU`` dilation (int): Spacing between kernel elements. Default: 1 inplace (bool): Parameter for the activation layer, which can optionally do the operation in-place. Default ``True`` bias (bool, optional): Whether to use bias in the convolution layer. By default, biases are included if ``norm_layer is None``. rIr,NTrJrKrLrMrNrOrP.rQrRrSrr*c jt |||||||||| | | tjj yN)rrrrerh rrJrKrLrMrNrOrPrQrRrSrrs rrzConv2dNormActivation.__init__>             HHOO r)r9r<r=r>rrerfrgr?rr]rrBrrirCrrErFs@rrkrk~s<,45./>B?Dxx?S?SEJXX]]01"&#   3c3h/0  c5c?*+  %U38_c 9:;   Xc588??&:;< #8C,@#AB U38_,- $ tn    rrkceZdZdZddddej j ej jdddf dedede ee eeeffd e ee eeeffd e e ee eeefe fd ed e e d ej jfde e d ej jfde ee eeeffde ede eddffd ZxZS)Conv3dNormActivationa Configurable block used for Convolution3d-Normalization-Activation blocks. Args: in_channels (int): Number of channels in the input video. out_channels (int): Number of channels produced by the Convolution-Normalization-Activation block kernel_size: (int, optional): Size of the convolving kernel. Default: 3 stride (int, optional): Stride of the convolution. Default: 1 padding (int, tuple or str, optional): Padding added to all four sides of the input. Default: None, in which case it will be calculated as ``padding = (kernel_size - 1) // 2 * dilation`` groups (int, optional): Number of blocked connections from input channels to output channels. Default: 1 norm_layer (Callable[..., torch.nn.Module], optional): Norm layer that will be stacked on top of the convolution layer. If ``None`` this layer won't be used. Default: ``torch.nn.BatchNorm3d`` activation_layer (Callable[..., torch.nn.Module], optional): Activation function which will be stacked on top of the normalization layer (if not None), otherwise on top of the conv layer. If ``None`` this layer won't be used. Default: ``torch.nn.ReLU`` dilation (int): Spacing between kernel elements. Default: 1 inplace (bool): Parameter for the activation layer, which can optionally do the operation in-place. Default ``True`` bias (bool, optional): Whether to use bias in the convolution layer. By default, biases are included if ``norm_layer is None``. rIr,NTrJrKrLrMrNrOrP.rQrRrSrr*c jt |||||||||| | | tjj yrm)rrrreConv3drns rrzConv3dNormActivation.__init__ror)r9r<r=r>rre BatchNorm3drgr?rr]rrBrrirCrrErFs@rrqrqsH*9:34CG?Dxx?S?SEJXX]]56"&#   3c3m 445  c5c3//0  %U3S=%93 >?@   Xc588??&:;< #8C,@#AB U3S=112 $ tn    rrqc eZdZdZej j ej jfdedede dej jfde dej jfddf fd Z d e de fd Z d e de fd ZxZS) SqueezeExcitationaE This block implements the Squeeze-and-Excitation block from https://arxiv.org/abs/1709.01507 (see Fig. 1). Parameters ``activation``, and ``scale_activation`` correspond to ``delta`` and ``sigma`` in eq. 3. Args: input_channels (int): Number of channels in the input image squeeze_channels (int): Number of squeeze channels activation (Callable[..., torch.nn.Module], optional): ``delta`` activation. Default: ``torch.nn.ReLU`` scale_activation (Callable[..., torch.nn.Module]): ``sigma`` activation. Default: ``torch.nn.Sigmoid`` input_channelssqueeze_channels activation.scale_activationr*NcHt|t|tjj d|_tjj||d|_tjj||d|_ ||_ ||_ y)Nr,) rrr rreAdaptiveAvgPool2davgpoolrhfc1fc2ryrz)rrwrxryrzrs rrzSqueezeExcitation.__init__st D!xx11!4 88??>3CQG88??#3^QG$, 0 2rinputc|j|}|j|}|j|}|j|}|j |Srm)r}r~ryrrzrrr4s r_scalezSqueezeExcitation._scalesI U#&$$U++rc.|j|}||zSrm)rrs rr5zSqueezeExcitation.forwards E"u}r)r9r<r=r>rrergSigmoidr?rrirrrr5rErFs@rrvrvs 6;XX]];@88;K;K 3 3 3S%((//12 3 #3#78 3  3,F,v,VrrvceZdZdZdej j dddfdedeede e dej jfd e e dej jfd e e d e d e ffd ZxZS)MLPaThis block implements the multi-layer perceptron (MLP) module. Args: in_channels (int): Number of channels of the input hidden_channels (List[int]): List of the hidden channel dimensions norm_layer (Callable[..., torch.nn.Module], optional): Norm layer that will be stacked on top of the linear layer. If ``None`` this layer won't be used. Default: ``None`` activation_layer (Callable[..., torch.nn.Module], optional): Activation function which will be stacked on top of the normalization layer (if not None), otherwise on top of the linear layer. If ``None`` this layer won't be used. Default: ``torch.nn.ReLU`` inplace (bool, optional): Parameter for the activation layer, which can optionally do the operation in-place. Default is ``None``, which uses the respective default values of the ``activation_layer`` and Dropout layer. bias (bool): Whether to use bias in the linear layer. Default ``True`` dropout (float): The probability for the dropout layer. Default: 0.0 NTgrJhidden_channelsrP.rQrSrdropoutcJ|ind|i}g} |} |ddD]} | jtjj| | ||| j|| | j|di|| jtjj|fi|| } | jtjj| |d|| jtjj|fi|t || t|y)NrSr-)rrW)r_rreLinearDropoutrrr ) rrJrrPrQrSrrrdrcin_dim hidden_dimrs rrz MLP.__init__sY,@)#2. J MM%((//&*4/H I% j45 MM*4V4 5 MM%((**7=f= >F    ehhoofob.AoMN ehh&&w9&9: &!D!r)r9r<r=r>rrergr?rDrrrirCr@rrErFs@rrrs "@DEJXX]]"&""c"Xc588??&:;< " #8C,@#AB " $ """"rrc>eZdZdZdeeffd ZdedefdZxZ S)PermutezThis module returns a view of the tensor input with its dimensions permuted. Args: dims (List[int]): The desired ordering of dimensions dimsc0t|||_yrm)rrr)rrrs rrzPermute.__init__<s  rr)r*cBtj||jSrm)rpermuter)rr)s rr5zPermute.forward@s}}Q **r) r9r<r=r>rDr?rrr5rErFs@rrr5s+ T#Y++F+rr)r`collections.abcrtypingrrrrrutilsr r re functional interpolaterir SequentialrHrkrqrvrrrWrrrs$,, 5hh!!-- 4T4Tn6,,6r/ -/ d. -. b$$N*"%((  *"Z +ehhoo +r