§
    ’©ºjÞD  ã                   óä  — d dl Z d dlZd dlmZmZmZmZ d dlZd dlmZm	Z	 d dl
mZ d dlmZ dddd	œZesd
dgZd>dedede	j        j        j        fd„Z G d„ dej	        j        ¦  «        Z G d„ dej	        j        ¦  «        Z G d„ dej	        j        ¦  «        Zd?dededefd„Zd@dedeedf         defd„ZdAd ed!edefd"„ZdBd$ed%edefd&„Z  G d'„ d(e	j        ¦  «        Z! G d)„ d*e	j        ¦  «        Z" G d+„ d,e	j        ¦  «        Z# G d-„ d.e#¦  «        Z$d/ed0edefd1„Z%d2ed3ede	j        d0edef
d4„Z&dCd6ed7eed8                  defd9„Z'	 	 	 dDd2ed3ed;ed<         d7eed8                  d0edefd=„Z(dS )Eé    N)ÚListÚ
NamedTupleÚOptionalÚUnion)ÚTensorÚnn)ÚLiteral)Ú_TORCHVISION_AVAILABLEÚSqueezeNet1_1_WeightsÚAlexNet_WeightsÚVGG16_Weights)Úsqueezenet1_1ÚalexnetÚvgg16Ú)learned_perceptual_image_patch_similarityÚ_get_tv_model_featuresFÚnetÚ
pretrainedÚreturnc                 ó  — t           st          d¦  «        ‚ddl}|rEt          |j        t
          |          ¦  «        } t          |j        | ¦  «        |j        ¬¦  «        }n t          |j        | ¦  «        d¬¦  «        }|j        S )aA  Get torchvision network.

    Args:
        net: Name of network
        pretrained: If pretrained weights should be used

    >>> _ = _get_tv_model_features("alexnet", pretrained=True)
    >>> _ = _get_tv_model_features("squeezenet1_1", pretrained=True)
    >>> _ = _get_tv_model_features("vgg16", pretrained=True)

    zSTorchvision is not installed. Please install torchvision to use this functionality.r   N)Úweights)r
   ÚModuleNotFoundErrorÚtorchvisionÚgetattrÚmodelsÚ_weight_mapÚDEFAULTÚfeatures)r   r   r   Úmodel_weightsÚmodels        ús/root/private/project/pallet_detection_v2/.venv/lib/python3.11/site-packages/torchmetrics/functional/image/lpips.pyr   r   -   s’   € õ "ð yÝ!Ð"wÑxÔxÐxØÐÐÐàð ?Ý Ô 2µKÀÔ4DÑEÔEˆØ0•˜Ô*¨CÑ0Ô0¸Ô9NÐOÑOÔOˆˆà0•˜Ô*¨CÑ0Ô0¸Ð>Ñ>Ô>ˆØŒ>Ðó    c                   óB   ‡ — e Zd ZdZddededdfˆ fd„Zd	edefd
„Zˆ xZ	S )Ú
SqueezeNetzSqueezeNet implementation.FTÚrequires_gradr   r   Nc           
      ó�  •— t          ¦   «                              ¦   «          t          d|¦  «        }d| _        g }t	          d¦  «        t	          dd¦  «        t	          dd¦  «        t	          dd¦  «        t	          dd¦  «        t	          dd¦  «        t	          dd	¦  «        g}|D ]c}t
          j                             ¦   «         }|D ]+}|                     t          |¦  «        ||         ¦  «         Œ,| 
                    |¦  «         Œdt          j        |¦  «        | _        |s|                      ¦   «         D ]}	d
|	_        Œ
d S d S )Nr   é   é   é   é   é
   é   é   é   F)ÚsuperÚ__init__r   ÚN_slicesÚrangeÚtorchr   Ú
SequentialÚ
add_moduleÚstrÚappendÚ
ModuleListÚslicesÚ
parametersr%   )Úselfr%   r   Úpretrained_featuresr9   Úfeature_rangesÚfeature_rangeÚseqÚiÚparamÚ	__class__s             €r!   r0   zSqueezeNet.__init__H   sE  ø€ Ý‰Œ×ÒÑÔÐÝ4°_ÀjÑQÔQÐàˆŒØˆÝ ™(œ(¥E¨!¨Q¡K¤Kµ°q¸!±´½eÀAÀr¹l¼lÍEÐRTÐVXÉMÌMÕ[`ÐacÐegÑ[hÔ[hÕjoÐprÐtvÑjwÔjwÐxˆØ+ð 	ð 	ˆMÝ”(×%Ò%Ñ'Ô'ˆCØ"ð ?ð ?�Ø—’�s 1™vœvÐ':¸1Ô'=Ñ>Ô>Ð>Ð>Ø�MŠM˜#ÑÔÐÐå”m FÑ+Ô+ˆŒØð 	,ØŸšÑ*Ô*ð ,ð ,�Ø&+�Ô#Ð#ð	,ð 	,ð,ð ,r"   Úxc                 óŠ   —  G d„ dt           ¦  «        }g }| j        D ]"} ||¦  «        }|                     |¦  «         Œ# ||Ž S )úProcess input.c                   óV   — e Zd ZU eed<   eed<   eed<   eed<   eed<   eed<   eed<   dS )	ú*SqueezeNet.forward.<locals>._SqueezeOutputÚrelu1Úrelu2Úrelu3Úrelu4Úrelu5Úrelu6Úrelu7N©Ú__name__Ú
__module__Ú__qualname__r   Ú__annotations__© r"   r!   Ú_SqueezeOutputrG   ]   sO   € € € € € € ØˆMˆM‰MØˆMˆM‰MØˆMˆM‰MØˆMˆM‰MØˆMˆM‰MØˆMˆM‰MØˆMˆM‰MˆMˆMr"   rU   )r   r9   r7   )r;   rC   rU   ÚrelusÚslice_s        r!   ÚforwardzSqueezeNet.forwardZ   so   € ð	ð 	ð 	ð 	ð 	�Zñ 	ô 	ð 	ð ˆØ”kð 	ð 	ˆFØ��q‘	”	ˆAØ�LŠL˜‰OŒOˆOˆOØˆ~˜uÐ%Ð%r"   ©FT©
rP   rQ   rR   Ú__doc__Úboolr0   r   r   rX   Ú__classcell__©rB   s   @r!   r$   r$   E   s€   ø€ € € € € Ø$Ð$ð,ð , dð ,Àð ,ÐPTð ,ð ,ð ,ð ,ð ,ð ,ð$&˜ð & Jð &ð &ð &ð &ð &ð &ð &ð &r"   r$   c                   óB   ‡ — e Zd ZdZddededdfˆ fd„Zd	edefd
„Zˆ xZ	S )ÚAlexnetzAlexnet implementation.FTr%   r   r   Nc                 ó   •— t          ¦   «                              ¦   «          t          d|¦  «        }t          j                             ¦   «         | _        t          j                             ¦   «         | _        t          j                             ¦   «         | _        t          j                             ¦   «         | _	        t          j                             ¦   «         | _
        d| _        t          d¦  «        D ]0}| j                             t          |¦  «        ||         ¦  «         Œ1t          dd¦  «        D ]0}| j                             t          |¦  «        ||         ¦  «         Œ1t          dd¦  «        D ]0}| j                             t          |¦  «        ||         ¦  «         Œ1t          dd¦  «        D ]0}| j	                             t          |¦  «        ||         ¦  «         Œ1t          dd¦  «        D ]0}| j
                             t          |¦  «        ||         ¦  «         Œ1|s|                      ¦   «         D ]}d|_        Œ
d S d S )Nr   r)   r(   r*   r+   r-   F©r/   r0   r   r3   r   r4   Úslice1Úslice2Úslice3Úslice4Úslice5r1   r2   r5   r6   r:   r%   )r;   r%   r   Úalexnet_pretrained_featuresrC   rA   rB   s         €r!   r0   zAlexnet.__init__p   s  ø€ Ý‰Œ×ÒÑÔÐÝ&<¸YÈ
Ñ&SÔ&SÐ#å”h×)Ò)Ñ+Ô+ˆŒÝ”h×)Ò)Ñ+Ô+ˆŒÝ”h×)Ò)Ñ+Ô+ˆŒÝ”h×)Ò)Ñ+Ô+ˆŒÝ”h×)Ò)Ñ+Ô+ˆŒØˆŒÝ�q‘”ð 	Kð 	KˆAØŒK×"Ò"¥3 q¡6¤6Ð+FÀqÔ+IÑJÔJÐJÐJÝ�q˜!‘”ð 	Kð 	KˆAØŒK×"Ò"¥3 q¡6¤6Ð+FÀqÔ+IÑJÔJÐJÐJÝ�q˜!‘”ð 	Kð 	KˆAØŒK×"Ò"¥3 q¡6¤6Ð+FÀqÔ+IÑJÔJÐJÐJÝ�q˜"‘”ð 	Kð 	KˆAØŒK×"Ò"¥3 q¡6¤6Ð+FÀqÔ+IÑJÔJÐJÐJÝ�r˜2‘”ð 	Kð 	KˆAØŒK×"Ò"¥3 q¡6¤6Ð+FÀqÔ+IÑJÔJÐJÐJØð 	,ØŸšÑ*Ô*ð ,ð ,�Ø&+�Ô#Ð#ð	,ð 	,ð,ð ,r"   rC   c                 ó,  — |                       |¦  «        }|}|                      |¦  «        }|}|                      |¦  «        }|}|                      |¦  «        }|}|                      |¦  «        }|} G d„ dt
          ¦  «        } ||||||¦  «        S )rE   c                   óB   — e Zd ZU eed<   eed<   eed<   eed<   eed<   dS )ú(Alexnet.forward.<locals>._AlexnetOutputsrH   rI   rJ   rK   rL   NrO   rT   r"   r!   Ú_AlexnetOutputsrk   •   s=   € € € € € € ØˆMˆM‰MØˆMˆM‰MØˆMˆM‰MØˆMˆM‰MØˆMˆM‰MˆMˆMr"   rl   ©rc   rd   re   rf   rg   r   )	r;   rC   ÚhÚh_relu1Úh_relu2Úh_relu3Úh_relu4Úh_relu5rl   s	            r!   rX   zAlexnet.forwardˆ   sª   € à�KŠK˜‰NŒNˆØˆØ�KŠK˜‰NŒNˆØˆØ�KŠK˜‰NŒNˆØˆØ�KŠK˜‰NŒNˆØˆØ�KŠK˜‰NŒNˆØˆð	ð 	ð 	ð 	ð 	�jñ 	ô 	ð 	ð ˆ˜w¨°¸'À7ÑKÔKÐKr"   rY   rZ   r^   s   @r!   r`   r`   m   sŠ   ø€ € € € € Ø!Ð!ð,ð , dð ,Àð ,ÐPTð ,ð ,ð ,ð ,ð ,ð ,ð0L˜ð L Jð Lð Lð Lð Lð Lð Lð Lð Lr"   r`   c                   óB   ‡ — e Zd ZdZddededdfˆ fd„Zd	edefd
„Zˆ xZ	S )ÚVgg16zVgg16 implementation.FTr%   r   r   Nc                 ó   •— t          ¦   «                              ¦   «          t          d|¦  «        }t          j                             ¦   «         | _        t          j                             ¦   «         | _        t          j                             ¦   «         | _        t          j                             ¦   «         | _	        t          j                             ¦   «         | _
        d| _        t          d¦  «        D ]0}| j                             t          |¦  «        ||         ¦  «         Œ1t          dd¦  «        D ]0}| j                             t          |¦  «        ||         ¦  «         Œ1t          dd¦  «        D ]0}| j                             t          |¦  «        ||         ¦  «         Œ1t          dd¦  «        D ]0}| j	                             t          |¦  «        ||         ¦  «         Œ1t          dd¦  «        D ]0}| j
                             t          |¦  «        ||         ¦  «         Œ1|s|                      ¦   «         D ]}d|_        Œ
d S d S )	Nr   r)   é   é	   é   é   é   Frb   )r;   r%   r   Úvgg_pretrained_featuresrC   rA   rB   s         €r!   r0   zVgg16.__init__¢   s  ø€ Ý‰Œ×ÒÑÔÐÝ"8¸À*Ñ"MÔ"MÐå”h×)Ò)Ñ+Ô+ˆŒÝ”h×)Ò)Ñ+Ô+ˆŒÝ”h×)Ò)Ñ+Ô+ˆŒÝ”h×)Ò)Ñ+Ô+ˆŒÝ”h×)Ò)Ñ+Ô+ˆŒØˆŒÝ�q‘”ð 	Gð 	GˆAØŒK×"Ò"¥3 q¡6¤6Ð+BÀ1Ô+EÑFÔFÐFÐFÝ�q˜!‘”ð 	Gð 	GˆAØŒK×"Ò"¥3 q¡6¤6Ð+BÀ1Ô+EÑFÔFÐFÐFÝ�q˜"‘”ð 	Gð 	GˆAØŒK×"Ò"¥3 q¡6¤6Ð+BÀ1Ô+EÑFÔFÐFÐFÝ�r˜2‘”ð 	Gð 	GˆAØŒK×"Ò"¥3 q¡6¤6Ð+BÀ1Ô+EÑFÔFÐFÐFÝ�r˜2‘”ð 	Gð 	GˆAØŒK×"Ò"¥3 q¡6¤6Ð+BÀ1Ô+EÑFÔFÐFÐFØð 	,ØŸšÑ*Ô*ð ,ð ,�Ø&+�Ô#Ð#ð	,ð 	,ð,ð ,r"   rC   c                 ó,  — |                       |¦  «        }|}|                      |¦  «        }|}|                      |¦  «        }|}|                      |¦  «        }|}|                      |¦  «        }|} G d„ dt
          ¦  «        } ||||||¦  «        S )rE   c                   óB   — e Zd ZU eed<   eed<   eed<   eed<   eed<   dS )ú"Vgg16.forward.<locals>._VGGOutputsÚrelu1_2Úrelu2_2Úrelu3_3Úrelu4_3Úrelu5_3NrO   rT   r"   r!   Ú_VGGOutputsr   Ç   s=   € € € € € € ØˆOˆO‰OØˆOˆO‰OØˆOˆO‰OØˆOˆO‰OØˆOˆO‰OˆOˆOr"   r…   rm   )	r;   rC   rn   Ú	h_relu1_2Ú	h_relu2_2Ú	h_relu3_3Ú	h_relu4_3Ú	h_relu5_3r…   s	            r!   rX   zVgg16.forwardº   sª   € à�KŠK˜‰NŒNˆØˆ	Ø�KŠK˜‰NŒNˆØˆ	Ø�KŠK˜‰NŒNˆØˆ	Ø�KŠK˜‰NŒNˆØˆ	Ø�KŠK˜‰NŒNˆØˆ	ð	ð 	ð 	ð 	ð 	�*ñ 	ô 	ð 	ð ˆ{˜9 i°¸IÀyÑQÔQÐQr"   rY   rZ   r^   s   @r!   ru   ru   Ÿ   sŠ   ø€ € € € € ØÐð,ð , dð ,Àð ,ÐPTð ,ð ,ð ,ð ,ð ,ð ,ð0R˜ð R Jð Rð Rð Rð Rð Rð Rð Rð Rr"   ru   TÚin_tensÚkeep_dimc                 ó4   — |                       ddg|¬¦  «        S )z2Spatial averaging over height and width of images.r(   é   )Úkeepdim©Úmean)r‹   rŒ   s     r!   Ú_spatial_averager’   Ñ   s   € à�<Š<˜˜A˜¨ˆ<Ñ1Ô1Ð1r"   ©é@   r”   Úout_hw.c                 óB   —  t          j        |dd¬¦  «        | ¦  «        S )z+Upsample input with bilinear interpolation.ÚbilinearF)ÚsizeÚmodeÚalign_corners)r   ÚUpsample)r‹   r•   s     r!   Ú	_upsamplerœ   Ö   s#   € àI�2Œ;˜F¨À5ÐIÑIÔIÈ'ÑRÔRÐRr"   ç:Œ0âŽyE>Úin_featÚepsc                 ój   — t          j        |t          j        | dz  dd¬¦  «        z   ¦  «        }| |z  S )zNormalize input tensor.r(   é   T)Údimr�   )r3   ÚsqrtÚsum)rž   rŸ   Únorm_factors      r!   Ú_normalize_tensorr¦   Û   s6   € å”*˜S¥5¤9¨W°a©Z¸QÈÐ#MÑ#MÔ#MÑMÑNÔN€KØ�[Ñ Ð r"   r”   rC   r˜   c                 óì   — | j         d         |k    r:| j         d         |k    r)t          j        j                             | ||fd¬¦  «        S t          j        j                             | ||fdd¬¦  «        S )zlhttps://github.com/toshas/torch-fidelity/blob/master/torch_fidelity/sample_similarity_lpips.py#L127C22-L132.éÿÿÿÿéþÿÿÿÚarea)r™   r—   F)r™   rš   )Úshaper3   r   Ú
functionalÚinterpolate)rC   r˜   s     r!   Ú_resize_tensorr®   á   sn   € à„wˆr„{�TÒÐ˜aœg bœk¨DÒ0Ð0ÝŒxÔ"×.Ò.¨q°4¸°,ÀVÐ.ÑLÔLÐLÝŒ8Ô×*Ò*¨1¨t°T¨lÀÐ[`Ð*ÑaÔaÐar"   c                   óJ   ‡ — e Zd ZU dZeed<   eed<   d	ˆ fd„Zdedefd„Zˆ xZS )
ÚScalingLayerzScaling layer.ÚshiftÚscaler   Nc                 ó(  •— t          ¦   «                              ¦   «          |                      dt          j        g d¢¦  «        d d d …d d f         d¬¦  «         |                      dt          j        g d¢¦  «        d d d …d d f         d¬¦  «         d S )Nr±   )g¸…ëQ¸ž¿gºI+‡¶¿gªñÒMbÈ¿F)Ú
persistentr²   )gZd;ßOÝ?gyé&1¬Ü?gÍÌÌÌÌÌÜ?)r/   r0   Úregister_bufferr3   r   )r;   rB   s    €r!   r0   zScalingLayer.__init__î   s§   ø€ Ý‰Œ×ÒÑÔÐØ×Ò˜W¥e¤lÐ3KÐ3KÐ3KÑ&LÔ&LÈTÐSTÐSTÐSTÐVZÐ\`ÐM`Ô&aÐnsÐÑtÔtÐtØ×Ò˜W¥e¤lÐ3HÐ3HÐ3HÑ&IÔ&IÈ$ÐPQÐPQÐPQÐSWÐY]ÐJ]Ô&^ÐkpÐÑqÔqÐqÐqÐqr"   Úinpc                 ó&   — || j         z
  | j        z  S ©rE   )r±   r²   )r;   r¶   s     r!   rX   zScalingLayer.forwardó   s   € à�d”jÑ  D¤JÑ.Ð.r"   )r   N)	rP   rQ   rR   r[   r   rS   r0   rX   r]   r^   s   @r!   r°   r°   è   s   ø€ € € € € € ØÐà€M€M�MØ€M€M�Mðrð rð rð rð rð rð
/˜6ð / fð /ð /ð /ð /ð /ð /ð /ð /r"   r°   c            	       óF   ‡ — e Zd ZdZddedededdfˆ fd	„Zd
edefd„Zˆ xZ	S )ÚNetLinLayerz,A single linear layer which does a 1x1 conv.r¡   FÚchn_inÚchn_outÚuse_dropoutr   Nc           	      óÚ   •— t          ¦   «                              ¦   «          |rt          j        ¦   «         gng }|t          j        ||dddd¬¦  «        gz  }t          j        |Ž | _        d S )Nr¡   r   F)ÚstrideÚpaddingÚbias)r/   r0   r   ÚDropoutÚConv2dr4   r    )r;   r»   r¼   r½   ÚlayersrB   s        €r!   r0   zNetLinLayer.__init__û   sl   ø€ Ý‰Œ×ÒÑÔÐà#.Ð6•"”*‘,”,��°BˆØÝŒI�f˜g q°¸AÀEÐJÑJÔJð
ñ 	
ˆõ ”] FÐ+ˆŒ
ˆ
ˆ
r"   rC   c                 ó,   — |                       |¦  «        S r¸   )r    )r;   rC   s     r!   rX   zNetLinLayer.forward  s   € à�zŠz˜!‰}Œ}Ðr"   )r¡   F)
rP   rQ   rR   r[   Úintr\   r0   r   rX   r]   r^   s   @r!   rº   rº   ø   s‡   ø€ € € € € Ø6Ð6ð,ð ,˜sð ,¨Sð ,À4ð ,ÐTXð ,ð ,ð ,ð ,ð ,ð ,ð˜ð  Fð ð ð ð ð ð ð ð r"   rº   c                   óÌ   ‡ — e Zd Z	 	 	 	 	 	 	 	 	 ddeded         ded	ed
ededee         dedee         ddfˆ fd„Z	 dde	de	dedede
e	ee	ee	         f         f         f
d„Zˆ xZS )Ú_LPIPSTÚalexFNr   r   ©rÉ   ÚvggÚsqueezeÚspatialÚ	pnet_randÚ	pnet_tuner½   Ú
model_pathÚ	eval_modeÚresizer   c
           	      óî  •— t          ¦   «                              ¦   «          || _        || _        || _        || _        |	| _        t          ¦   «         | _        | j        dv rt          }
g d¢| _
        n7| j        dk    rt          }
g d¢| _
        n| j        dk    rt          }
g d¢| _
        t          | j
        ¦  «        | _         |
| j         | j        ¬¦  «        | _        t!          | j
        d         |¬	¦  «        | _        t!          | j
        d
         |¬	¦  «        | _        t!          | j
        d         |¬	¦  «        | _        t!          | j
        d         |¬	¦  «        | _        t!          | j
        d         |¬	¦  «        | _        | j        | j        | j        | j        | j        g| _        | j        dk    r^t!          | j
        d         |¬	¦  «        | _        t!          | j
        d         |¬	¦  «        | _        | xj        | j        | j        gz  c_        t3          j        | j        ¦  «        | _        |r†|€Yt6          j                             t6          j                             t?          j         | j        ¦  «        dd|› d�¦  «        ¦  «        }|  !                    tE          j#        |d¬¦  «        d¬¦  «         |r|  $                    ¦   «          | j        s|  %                    ¦   «         D ]}d|_&        Œ
dS dS )aß  Initializes a perceptual loss torch.nn.Module.

        Args:
            pretrained: This flag controls the linear layers should be pretrained version or random
            net: Indicate backbone to use, choose between ['alex','vgg','squeeze']
            spatial: If input should be spatial averaged
            pnet_rand: If backbone should be random or use imagenet pre-trained weights
            pnet_tune: If backprop should be enabled for both backbone and linear layers
            use_dropout: If dropout layers should be added
            model_path: Model path to load pretained models from
            eval_mode: If network should be in evaluation mode
            resize: If input should be resized to this size

        )rË   r   )r”   é€   é   é   rÖ   rÉ   )r”   éÀ   é€  rÕ   rÕ   rÌ   )r”   rÔ   rÕ   rØ   rØ   rÖ   rÖ   )r   r%   r   )r½   r¡   r(   rŽ   rw   r)   é   Nz..zlpips_models/z.pthÚcpu)Úmap_locationF)Ústrict)'r/   r0   Ú	pnet_typerÏ   rÎ   rÍ   rÒ   r°   Úscaling_layerru   Úchnsr`   r$   ÚlenÚLr   rº   Úlin0Úlin1Úlin2Úlin3Úlin4ÚlinsÚlin5Úlin6r   r8   ÚosÚpathÚabspathÚjoinÚinspectÚgetfileÚload_state_dictr3   ÚloadÚevalr:   r%   )r;   r   r   rÍ   rÎ   rÏ   r½   rÐ   rÑ   rÒ   Únet_typerA   rB   s               €r!   r0   z_LPIPS.__init__
  s·  ø€ õ4 	‰Œ×ÒÑÔÐàˆŒØ"ˆŒØ"ˆŒØˆŒØˆŒÝ)™^œ^ˆÔàŒ>Ð-Ð-Ð-ÝˆHØ0Ð0Ð0ˆDŒIˆIØŒ^˜vÒ%Ð%ÝˆHØ0Ð0Ð0ˆDŒIˆIØŒ^˜yÒ(Ð(Ý!ˆHØ:Ð:Ð:ˆDŒIÝ�T”Y‘”ˆŒà�8¨4¬>Ð'9ÈÌÐXÑXÔXˆŒå ¤	¨!¤¸+ÐFÑFÔFˆŒ	Ý ¤	¨!¤¸+ÐFÑFÔFˆŒ	Ý ¤	¨!¤¸+ÐFÑFÔFˆŒ	Ý ¤	¨!¤¸+ÐFÑFÔFˆŒ	Ý ¤	¨!¤¸+ÐFÑFÔFˆŒ	Ø”Y ¤	¨4¬9°d´iÀÄÐKˆŒ	ØŒ>˜YÒ&Ð&Ý# D¤I¨a¤L¸kÐJÑJÔJˆDŒIÝ# D¤I¨a¤L¸kÐJÑJÔJˆDŒIØˆIŒI˜$œ) T¤YÐ/Ñ/ˆIŒIÝ”M $¤)Ñ,Ô,ˆŒ	àð 	[ØÐ!ÝœWŸ_š_Ý”G—L’L¥¤°´Ñ!?Ô!?ÀÐG`ÐWZÐG`ÐG`ÐG`ÑaÔañô �
ð × Ò ¥¤¨JÀUÐ!KÑ!KÔ!KÐTYÐ ÑZÔZÐZàð 	Ø�IŠI‰KŒKˆKàŒ~ð 	,ØŸšÑ*Ô*ð ,ð ,�Ø&+�Ô#Ð#ð	,ð 	,ð,ð ,r"   Úin0Úin1ÚretperlayerÚ	normalizec                 ó®  — |rd|z  dz
  }d|z  dz
  }|                       |¦  «        |                       |¦  «        }}| j        �,t          || j        ¬¦  «        }t          || j        ¬¦  «        }| j                             |¦  «        | j                             |¦  «        }}i i i }}
}	t          | j        ¦  «        D ]J}t          ||         ¦  «        t          ||         ¦  «        c|	|<   |
|<   |	|         |
|         z
  dz  ||<   ŒKg }t          | j        ¦  «        D ] }| j        rY| 	                    t           | j        |         ||         ¦  «        t          |j        dd …         ¦  «        ¬¦  «        ¦  «         Œb| 	                    t           | j        |         ||         ¦  «        d¬¦  «        ¦  «         Œ¡t          |¦  «        }|r||fS |S )Nr(   r¡   )r˜   )r•   T)rŒ   )rÞ   rÒ   r®   r   rX   r2   rá   r¦   rÍ   r7   rœ   rç   Útupler«   r’   r¤   )r;   rô   rõ   rö   r÷   Ú	in0_inputÚ	in1_inputÚouts0Úouts1Úfeats0Úfeats1ÚdiffsÚkkÚresÚvals                  r!   rX   z_LPIPS.forwardU  sæ  € ð ð 	Ø�c‘'˜A‘+ˆCØ�c‘'˜A‘+ˆCð  $×1Ò1°#Ñ6Ô6¸×8JÒ8JÈ3Ñ8OÔ8O�9ˆ	ð Œ;Ð"Ý& y°t´{ÐCÑCÔCˆIÝ& y°t´{ÐCÑCÔCˆIà”x×'Ò'¨	Ñ2Ô2°D´H×4DÒ4DÀYÑ4OÔ4OˆuˆØ " B¨˜�ˆå˜œ‘-”-ð 	7ð 	7ˆBÝ%6°u¸R´yÑ%AÔ%AÕCTÐUZÐ[]ÔU^ÑC_ÔC_Ð"ˆF�2‰J˜˜r™
Ø œ f¨R¤jÑ0°QÑ6ˆE�"‰IˆIàˆÝ˜œ‘-”-ð 	Vð 	VˆBØŒ|ð VØ—
’
�9 ] T¤Y¨r¤]°5¸´9Ñ%=Ô%=ÅeÈCÌIÐVWÐVXÐVXÌMÑFZÔFZÐ[Ñ[Ô[Ñ\Ô\Ð\Ð\à—
’
Õ+¨M¨D¬I°b¬M¸%À¼)Ñ,DÔ,DÈtÐTÑTÔTÑUÔUÐUÐUå˜#‘h”hˆØð 	Ø˜�:ÐØˆ
r"   )	TrÉ   FFFTNTN)FF)rP   rQ   rR   r\   r	   r   r6   rÆ   r0   r   r   rù   r   rX   r]   r^   s   @r!   rÈ   rÈ   	  sA  ø€ € € € € ð  Ø17ØØØØ Ø$(ØØ $ðI,ð I,àðI,ð Ð-Ô.ðI,ð ð	I,ð
 ðI,ð ðI,ð ðI,ð ˜S”MðI,ð ðI,ð ˜”ðI,ð 
ðI,ð I,ð I,ð I,ð I,ð I,ðX V[ð ð  Øð Ø &ð Ø59ð ØNRð à	ˆv�u˜V T¨&¤\Ð1Ô2Ð2Ô	3ð ð  ð  ð  ð  ð  ð  ð  r"   rÈ   c                   ó,   ‡ — e Zd ZdZdedd fˆ fd„Zˆ xZS )Ú_NoTrainLpipsz8Wrapper to make sure LPIPS never leaves evaluation mode.r™   r   c                 óF   •— t          ¦   «                              d¦  «        S )z.Force network to always be in evaluation mode.F)r/   Útrain)r;   r™   rB   s     €r!   r  z_NoTrainLpips.train{  s   ø€ å‰wŒw�}Š}˜UÑ#Ô#Ð#r"   )rP   rQ   rR   r[   r\   r  r]   r^   s   @r!   r  r  x  sQ   ø€ € € € € ØBÐBð$˜$ð $ ?ð $ð $ð $ð $ð $ð $ð $ð $ð $ð $r"   r  Úimgr÷   c                 óÒ   — |r0|                       ¦   «         dk    o|                      ¦   «         dk    n|                      ¦   «         dk    }| j        dk    o| j        d         dk    o|S )z1Check that input is a valid image to the network.g      ð?g        r¨   rw   r¡   rŽ   )ÚmaxÚminÚndimr«   )r  r÷   Úvalue_checks      r!   Ú
_valid_imgr  €  s\   € à;DÐY�#—'’'‘)”)˜sÒ"Ð7 s§w¢w¡y¤y°CÒ'7øÈ#Ï'Ê'É)Ì)ÐWYÊ/€KØŒ8�qŠ=Ð>˜SœY qœ\¨QÒ.Ð>°;Ð>r"   Úimg1Úimg2c                 ó|  — t          | |¦  «        rt          ||¦  «        s}t          d| j        › d|j        › d|                      ¦   «         |                      ¦   «         g› d|                     ¦   «         |                     ¦   «         g› d|rddgnddg› d�¦  «        ‚ || ||¬	¦  «                             ¦   «         S )
NzeExpected both input arguments to be normalized tensors with shape [N, 3, H, W]. Got input with shape z and z and values in range z+ when all values are expected to be in the r   r¡   r¨   z range.)r÷   )r  Ú
ValueErrorr«   r  r
  rÌ   )r  r  r   r÷   s       r!   Ú_lpips_updater  †  sô   € Ý�t˜YÑ'Ô'ð 
­J°t¸YÑ,GÔ,Gð 
ÝðPØ%)¤ZðPð PØ6:´jðPð Pà—’‘”˜TŸXšX™ZœZÐ(ðPð Pà04·²±
´
¸D¿HºH¹J¼JÐ/GðPð Pð 1:Ð&F q¨! f fÀÀA¸wðPð Pð Pñ
ô 
ð 	
ð ˆ3ˆt�T YÐ/Ñ/Ô/×7Ò7Ñ9Ô9Ð9r"   r‘   ÚscoresÚ	reduction)r¤   r‘   Únonec                 ó¢   — |dk    r|                       ¦   «         S |dk    r|                      ¦   «         S |dk    s|€| S t          d|› �¦  «        ‚)Nr‘   r¤   r  zInvalid reduction type: )r‘   r¤   r  )r  r  s     r!   Ú_lpips_computer  ‘  s`   € Ø�FÒÐØ�{Š{‰}Œ}ÐØ�EÒÐØ�zŠz‰|Œ|ÐØ�FÒÐ˜iÐ/ØˆÝ
Ð;°	Ð;Ð;Ñ
<Ô
<Ð<r"   rÉ   ró   rÊ   c                 ó¤   — t          |¬¦  «                             | j        | j        ¬¦  «        }t	          | |||¦  «        }t          ||¦  «        S )a[  The Learned Perceptual Image Patch Similarity (`LPIPS_`) calculates perceptual similarity between two images.

    LPIPS essentially computes the similarity between the activations of two image patches for some pre-defined network.
    This measure has been shown to match human perception well. A low LPIPS score means that image patches are
    perceptual similar.

    Both input image patches are expected to have shape ``(N, 3, H, W)``. The minimum size of `H, W` depends on the
    chosen backbone (see `net_type` arg).

    Args:
        img1: first set of images
        img2: second set of images
        net_type: str indicating backbone network type to use. Choose between `'alex'`, `'vgg'` or `'squeeze'`
        reduction: str indicating how to reduce over the batch dimension. Choose between `'sum'`, `'mean'`, `'none'`
            or `None`.
        normalize: by default this is ``False`` meaning that the input is expected to be in the [-1,1] range. If set
            to ``True`` will instead expect input to be in the ``[0,1]`` range.

    Example:
        >>> from torch import rand
        >>> from torchmetrics.functional.image.lpips import learned_perceptual_image_patch_similarity
        >>> img1 = (rand(10, 3, 100, 100) * 2) - 1
        >>> img2 = (rand(10, 3, 100, 100) * 2) - 1
        >>> learned_perceptual_image_patch_similarity(img1, img2, net_type='squeeze')
        tensor(0.1005)

        >>> from torch import rand, Generator
        >>> from torchmetrics.functional.image.lpips import learned_perceptual_image_patch_similarity
        >>> gen = Generator().manual_seed(42)
        >>> img1 = (rand(2, 3, 100, 100, generator=gen) * 2) - 1
        >>> img2 = (rand(2, 3, 100, 100, generator=gen) * 2) - 1
        >>> learned_perceptual_image_patch_similarity(img1, img2, net_type='squeeze', reduction='none')
        tensor([0.1024, 0.0938])

    )r   )ÚdeviceÚdtype)r  Útor  r  r  r  )r  r  ró   r  r÷   r   Úlosss          r!   r   r   ›  sO   € õT ˜HÐ
%Ñ
%Ô
%×
(Ò
(°´À4Ä:Ð
(Ñ
NÔ
N€CÝ˜˜t S¨)Ñ4Ô4€DÝ˜$ 	Ñ*Ô*Ð*r"   )F)T)r“   )r�   )r”   r�   )rÉ   r‘   F))rî   rê   Útypingr   r   r   r   r3   r   r   Útyping_extensionsr	   Útorchmetrics.utilities.importsr
   r   Ú__doctest_skip__r6   r\   ÚmodulesÚ	containerr4   r   ÚModuler$   r`   ru   r’   rù   rÆ   rœ   Úfloatr¦   r®   r°   rº   rÈ   r  r  r  r  r   rT   r"   r!   ú<module>r&     s  ðð2 €€€Ø 	€	€	€	Ø 4Ð 4Ð 4Ð 4Ð 4Ð 4Ð 4Ð 4Ð 4Ð 4Ð 4Ð 4à €€€Ø Ð Ð Ð Ð Ð Ð Ð Ø %Ð %Ð %Ð %Ð %Ð %à AÐ AÐ AÐ AÐ AÐ Að -Ø Øðð €ð ð _ØCÐE]Ð^Ððð  ð °ð À"Ä*ÔBVÔBað ð ð ð ð0%&ð %&ð %&ð %&ð %&�””ñ %&ô %&ð %&ðP/Lð /Lð /Lð /Lð /LˆeŒhŒoñ /Lô /Lð /Lðd/Rð /Rð /Rð /Rð /RˆEŒHŒOñ /Rô /Rð /Rðd2ð 2˜fð 2°ð 2Àð 2ð 2ð 2ð 2ð
Sð S�vð S u¨S°#¨X¤ð SÀfð Sð Sð Sð Sð
!ð !˜vð !¨Eð !¸Vð !ð !ð !ð !ðbð b�fð b Cð b°ð bð bð bð bð/ð /ð /ð /ð /�2”9ñ /ô /ð /ð ð ð ð ð �"”)ñ ô ð ð"lð lð lð lð lˆRŒYñ lô lð lð^$ð $ð $ð $ð $�Fñ $ô $ð $ð?�Fð ? tð ?°ð ?ð ?ð ?ð ?ð:˜ð : fð :°2´9ð :Èð :ÐRXð :ð :ð :ð :ð=ð =˜6ð =¨h°wÐ?TÔ7UÔ.Vð =Ðdjð =ð =ð =ð =ð 39Ø:@Øð,+ð ,+Ø
ð,+à
ð,+ð Ð.Ô/ð,+ð ˜Ð 5Ô6Ô7ð	,+ð
 ð,+ð ð,+ð ,+ð ,+ð ,+ð ,+ð ,+r"   