§
    ‚Štj=  ã            
       ó´  — d Z ddlZddlmZ ddlZddlmZ ddlmZ ddlm	Z	 ddl
mZ dd	lmZ dd
lmZmZmZ ddlmZ  ej        e¦  «        Zdeej                 deej                 deej                 fd„Zdej        dededej        fd„Zdej        dej        dej        dedej        f
d„Zdej        dej        dedej        fd„Zdedej        fd„Z ed ¬!¦  «        e G d"„ d#e¦  «        ¦   «         ¦   «         Z  G d$„ d%ej!        ¦  «        Z" G d&„ d'ej!        ¦  «        Z# G d(„ d)ej!        ¦  «        Z$ G d*„ d+ej!        ¦  «        Z%d,e$iZ& G d-„ d.ej!        ¦  «        Z' G d/„ d0ej!        ¦  «        Z( G d1„ d2ej!        ¦  «        Z) G d3„ d4ej!        ¦  «        Z*e G d5„ d6e¦  «        ¦   «         Z+ ed7¬!¦  «         G d8„ d9e+¦  «        ¦   «         Z,d6d9gZ-dS ):zPyTorch SuperGlue model.é    N)Ú	dataclass)Únn)ÚPreTrainedModel)ÚSuperGlueConfigé   )Úinitialization)Úcreate_bidirectional_mask)ÚModelOutputÚauto_docstringÚloggingé   )ÚAutoModelForKeypointDetectionÚtensor_tuple0Útensor_tuple1Úreturnc                 óP   — t          d„ t          | |¦  «        D ¦   «         ¦  «        S )a'  
    Concatenate two tuples of tensors pairwise

    Args:
        tensor_tuple0 (`tuple[torch.Tensor]`):
            Tuple of tensors.
        tensor_tuple1 (`tuple[torch.Tensor]`):
            Tuple of tensors.

    Returns:
        (`tuple[torch.Tensor]`): Tuple of concatenated tensors.
    c              3   óH   K  — | ]\  }}t          j        ||g¦  «        V — Œd S ©N)ÚtorchÚcat)Ú.0Útensor0Útensor1s      ún/var/www/html/CA-Chatbot/venv/lib/python3.11/site-packages/transformers/models/superglue/modeling_superglue.pyú	<genexpr>zconcat_pairs.<locals>.<genexpr>/   s6   è è € ÐiÐiÑ3C°7¸G•”˜G WÐ-Ñ.Ô.ÐiÐiÐiÐiÐiÐió    )ÚtupleÚzip)r   r   s     r   Úconcat_pairsr   "   s,   € õ ÐiÐiÅsÈ=ÐZgÑGhÔGhÐiÑiÔiÑiÔiÐir   Ú	keypointsÚheightÚwidthc                 óà   — t          j        ||g| j        | j        ¬¦  «        d         }|dz  }|                     dd¬¦  «        j        dz  }| |dd…ddd…f         z
  |dd…ddd…f         z  S )a©  
    Normalize keypoints locations based on image image_shape

    Args:
        keypoints (`torch.Tensor` of shape `(batch_size, num_keypoints, 2)`):
            Keypoints locations in (x, y) format.
        height (`int`):
            Image height.
        width (`int`):
            Image width.

    Returns:
        Normalized keypoints locations of shape (`torch.Tensor` of shape `(batch_size, num_keypoints, 2)`).
    )ÚdeviceÚdtypeNr   é   T)Úkeepdimgffffffæ?)r   Útensorr$   r%   ÚmaxÚvalues)r    r!   r"   ÚsizeÚcenterÚscalings         r   Únormalize_keypointsr.   2   s…   € õ Œ<˜ ˜°	Ô0@È	ÌÐXÑXÔXÐY]Ô^€DØ�A‰X€FØ�hŠh�q $ˆhÑ'Ô'Ô.°Ñ4€GØ˜˜q˜q˜q $¨¨¨˜zÔ*Ñ*¨g°a°a°a¸¸q¸q¸q°jÔ.AÑAÐAr   Úlog_cost_matrixÚlog_source_distributionÚlog_target_distributionÚnum_iterationsc                 óŽ  — t          j        |¦  «        }t          j        |¦  «        }t          |¦  «        D ]`}|t          j        | |                     d¦  «        z   d¬¦  «        z
  }|t          j        | |                     d¦  «        z   d¬¦  «        z
  }Œa| |                     d¦  «        z   |                     d¦  «        z   S )az  
    Perform Sinkhorn Normalization in Log-space for stability

    Args:
        log_cost_matrix (`torch.Tensor` of shape `(batch_size, num_rows, num_columns)`):
            Logarithm of the cost matrix.
        log_source_distribution (`torch.Tensor` of shape `(batch_size, num_rows)`):
            Logarithm of the source distribution.
        log_target_distribution (`torch.Tensor` of shape `(batch_size, num_columns)`):
            Logarithm of the target distribution.

    Returns:
        log_cost_matrix (`torch.Tensor` of shape `(batch_size, num_rows, num_columns)`): Logarithm of the optimal
        transport matrix.
    r&   r   ©Údim)r   Ú
zeros_likeÚrangeÚ	logsumexpÚ	unsqueeze)r/   r0   r1   r2   Úlog_u_scalingÚlog_v_scalingÚ_s          r   Úlog_sinkhorn_iterationsr=   G   sÌ   € õ* Ô$Ð%<Ñ=Ô=€MÝÔ$Ð%<Ñ=Ô=€MÝ�>Ñ"Ô"ð wð wˆØ/µ%´/À/ÐTa×TkÒTkÐlmÑTnÔTnÑBnÐtuÐ2vÑ2vÔ2vÑvˆØ/µ%´/À/ÐTa×TkÒTkÐlmÑTnÔTnÑBnÐtuÐ2vÑ2vÔ2vÑvˆˆØ˜]×4Ò4°QÑ7Ô7Ñ7¸-×:QÒ:QÐRSÑ:TÔ:TÑTÐTr   ÚscoresÚ	reg_paramÚ
iterationsc                 ó’  — | j         \  }}}|                      d¦  «        }||z                       | ¦  «        ||z                       | ¦  «        }}|                     ||d¦  «        }	|                     |d|¦  «        }
|                     |dd¦  «        }t	          j        t	          j        | |	gd¦  «        t	          j        |
|gd¦  «        gd¦  «        }||z                        ¦   «          }t	          j        |                     |¦  «        |                     ¦   «         d         |z   g¦  «        }t	          j        |                     |¦  «        |                     ¦   «         d         |z   g¦  «        }|d                              |d¦  «        |d                              |d¦  «        }}t          ||||¬¦  «        }||z
  }|S )a  
    Perform Differentiable Optimal Transport in Log-space for stability

    Args:
        scores: (`torch.Tensor` of shape `(batch_size, num_rows, num_columns)`):
            Cost matrix.
        reg_param: (`torch.Tensor` of shape `(batch_size, 1, 1)`):
            Regularization parameter.
        iterations: (`int`):
            Number of Sinkhorn iterations.

    Returns:
        log_optimal_transport_matrix: (`torch.Tensor` of shape `(batch_size, num_rows, num_columns)`): Logarithm of the
        optimal transport matrix.
    r&   éÿÿÿÿN)r2   )ÚshapeÚ
new_tensorÚtoÚexpandr   r   Úlogr=   )r>   r?   r@   Ú
batch_sizeÚnum_rowsÚnum_columnsÚ
one_tensorÚnum_rows_tensorÚnum_columns_tensorÚsource_reg_paramÚtarget_reg_paramÚ	couplingsÚlog_normalizationr0   r1   Úlog_optimal_transport_matrixs                   r   Úlog_optimal_transportrS   d   sé  € ð  )/¬Ñ%€J�˜+Ø×"Ò" 1Ñ%Ô%€JØ+3°jÑ+@×*DÒ*DÀVÑ*LÔ*LÈ{Ð]gÑOg×NkÒNkÐlrÑNsÔNsÐ'€Oà ×'Ò'¨
°H¸aÑ@Ô@ÐØ ×'Ò'¨
°A°{ÑCÔCÐØ× Ò  ¨Q°Ñ2Ô2€Iå”	�5œ9 fÐ.>Ð%?ÀÑDÔDÅeÄiÐQaÐclÐPmÐoqÑFrÔFrÐsÐuvÑwÔw€Ià)Ð,>Ñ>×CÒCÑEÔEÐEÐÝ#œiØ	×	!Ò	! (Ñ	+Ô	+Ð-?×-CÒ-CÑ-EÔ-EÀdÔ-KÐN_Ñ-_Ð`ñô Ðõ $œiØ	×	!Ò	! +Ñ	.Ô	.°×0CÒ0CÑ0EÔ0EÀdÔ0KÐN_Ñ0_Ð`ñô Ðð 	  Ô%×,Ò,¨Z¸Ñ<Ô<Ø Ô%×,Ò,¨Z¸Ñ<Ô<ð 5Ðõ
 $;ØÐ*Ð,CÐT^ð$ñ $ô $Ð ð $@ÐBSÑ#SÐ Ø'Ð'r   r5   c                 ón   — |                       | j        |         ¦  «                             d¦  «        dz
  S )Nr   r&   )Únew_onesrC   Úcumsum)Úxr5   s     r   Úarange_likerX   ‘   s-   € Ø�:Š:�a”g˜c”lÑ#Ô#×*Ò*¨1Ñ-Ô-°Ñ1Ð1r   aò  
    Base class for outputs of SuperGlue keypoint matching models. Due to the nature of keypoint detection and matching, the number
    of keypoints is not fixed and can vary from image to image, which makes batching non-trivial. In the batch of
    images, the maximum number of matches is set as the dimension of the matches and matching scores. The mask tensor is
    used to indicate which values in the keypoints, matches and matching_scores tensors are keypoint matching
    information.
    )Úcustom_introc                   óþ   — e Zd ZU dZdZej        dz  ed<   dZej        dz  ed<   dZ	ej        dz  ed<   dZ
ej        dz  ed<   dZej        dz  ed<   dZeej                 dz  ed<   dZeej                 dz  ed	<   dS )
ÚSuperGlueKeypointMatchingOutputaú  
    loss (`torch.FloatTensor` of shape `(1,)`, *optional*):
        Loss computed during training.
    matches (`torch.FloatTensor` of shape `(batch_size, 2, num_matches)`):
        Index of keypoint matched in the other image.
    matching_scores (`torch.FloatTensor` of shape `(batch_size, 2, num_matches)`):
        Scores of predicted matches.
    keypoints (`torch.FloatTensor` of shape `(batch_size, num_keypoints, 2)`):
        Absolute (x, y) coordinates of predicted keypoints in a given image.
    mask (`torch.IntTensor` of shape `(batch_size, num_keypoints)`):
        Mask indicating which values in matches and matching_scores are keypoint matching information.
    hidden_states (`tuple[torch.FloatTensor, ...]`, *optional*):
        Tuple of `torch.FloatTensor` (one for the output of each stage) of shape `(batch_size, 2, num_channels,
        num_keypoints)`, returned when `output_hidden_states=True` is passed or when
        `config.output_hidden_states=True`)
    attentions (`tuple[torch.FloatTensor, ...]`, *optional*):
        Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, 2, num_heads, num_keypoints,
        num_keypoints)`, returned when `output_attentions=True` is passed or when `config.output_attentions=True`)
    NÚlossÚmatchesÚmatching_scoresr    ÚmaskÚhidden_statesÚ
attentions)Ú__name__Ú
__module__Ú__qualname__Ú__doc__r\   r   ÚFloatTensorÚ__annotations__r]   r^   r    r_   Ú	IntTensorr`   r   ra   © r   r   r[   r[   •   sÏ   € € € € € € ðð ð( &*€Dˆ%Ô
˜dÑ
"Ð)Ð)Ñ)Ø(,€GˆUÔ Ñ%Ð,Ð,Ñ,Ø04€O�UÔ&¨Ñ-Ð4Ð4Ñ4Ø*.€IˆuÔ  4Ñ'Ð.Ð.Ñ.Ø#'€Dˆ%Œ/˜DÑ
 Ð'Ð'Ñ'Ø59€M�5˜Ô*Ô+¨dÑ2Ð9Ð9Ñ9Ø26€J��eÔ'Ô(¨4Ñ/Ð6Ð6Ñ6Ð6Ð6r   r[   c                   óT   ‡ — e Zd Zdedededdfˆ fd„Zdej        dej        fd„Zˆ xZ	S )	ÚSuperGlueMultiLayerPerceptronÚconfigÚin_channelsÚout_channelsr   Nc                 óÞ   •— t          ¦   «                              ¦   «          t          j        ||¦  «        | _        t          j        |¦  «        | _        t          j        ¦   «         | _        d S r   )	ÚsuperÚ__init__r   ÚLinearÚlinearÚBatchNorm1dÚ
batch_normÚReLUÚ
activation)Úselfrl   rm   rn   Ú	__class__s       €r   rq   z&SuperGlueMultiLayerPerceptron.__init__¾   sN   ø€ Ý‰Œ×ÒÑÔÐÝ”i ¨\Ñ:Ô:ˆŒÝœ.¨Ñ6Ô6ˆŒÝœ'™)œ)ˆŒˆˆr   Úhidden_statec                 óÜ   — |                       |¦  «        }|                     dd¦  «        }|                      |¦  «        }|                     dd¦  «        }|                      |¦  «        }|S )NrB   éþÿÿÿ)rs   Ú	transposeru   rw   )rx   rz   s     r   Úforwardz%SuperGlueMultiLayerPerceptron.forwardÄ   se   € Ø—{’{ <Ñ0Ô0ˆØ#×-Ò-¨b°"Ñ5Ô5ˆØ—’ |Ñ4Ô4ˆØ#×-Ò-¨b°"Ñ5Ô5ˆØ—’ |Ñ4Ô4ˆØÐr   )
rb   rc   rd   r   Úintrq   r   ÚTensorr~   Ú__classcell__©ry   s   @r   rk   rk   ½   s�   ø€ € € € € ð$˜ð $¸Sð $ÐPSð $ÐX\ð $ð $ð $ð $ð $ð $ð E¤Lð °U´\ð ð ð ð ð ð ð ð r   rk   c                   ó”   ‡ — e Zd Zdeddfˆ fd„Z	 d
dej        dej        dedz  deej        eej                 dz  f         fd	„Z	ˆ xZ
S )ÚSuperGlueKeypointEncoderrl   r   Nc                 óp  •‡‡— t          ¦   «                              ¦   «          ‰j        }‰j        }dg|z   |gz   Šˆˆfd„t	          dt          ‰¦  «        dz
  ¦  «        D ¦   «         }|                     t          j        ‰d         ‰d         ¦  «        ¦  «         t          j	        |¦  «        | _
        d S )Nr   c                 óP   •— g | ]"}t          ‰‰|d z
           ‰|         ¦  «        ‘Œ#S ©r&   ©rk   )r   Úirl   Úencoder_channelss     €€r   ú
<listcomp>z5SuperGlueKeypointEncoder.__init__.<locals>.<listcomp>Õ   sF   ø€ ð 
ð 
ð 
àõ *¨&Ð2BÀ1ÀqÁ5Ô2IÐK[Ð\]ÔK^Ñ_Ô_ð
ð 
ð 
r   r&   r|   rB   )rp   rq   Úkeypoint_encoder_sizesÚhidden_sizer7   ÚlenÚappendr   rr   Ú
ModuleListÚencoder)rx   rl   Úlayer_sizesr�   ÚlayersrŠ   ry   s    `   @€r   rq   z!SuperGlueKeypointEncoder.__init__Î   sÅ   øøø€ Ý‰Œ×ÒÑÔÐØÔ3ˆØÔ(ˆà˜3 Ñ,°¨}Ñ<Ðð
ð 
ð 
ð 
ð 
å˜1�cÐ"2Ñ3Ô3°aÑ7Ñ8Ô8ð
ñ 
ô 
ˆð 	�Š•b”iÐ 0°Ô 4Ð6FÀrÔ6JÑKÔKÑLÔLÐLÝ”} VÑ,Ô,ˆŒˆˆr   Fr    r>   Úoutput_hidden_statesc                 óª   — |                      d¦  «        }t          j        ||gd¬¦  «        }|rdnd }| j        D ]} ||¦  «        }|r||fz   }Œ||fS )Nr   r4   ri   )r9   r   r   r‘   )rx   r    r>   r”   rz   Úall_hidden_statesÚlayers          r   r~   z SuperGlueKeypointEncoder.forwardÜ   s„   € ð ×!Ò! !Ñ$Ô$ˆÝ”y )¨VÐ!4¸!Ð<Ñ<Ô<ˆØ"6Ð@˜B˜B¸DÐØ”\ð 	Hð 	HˆEØ ˜5 Ñ.Ô.ˆLØ#ð HØ$5¸¸Ñ$GÐ!øØÐ.Ð.Ð.r   )F©rb   rc   rd   r   rq   r   r€   Úboolr   r~   r�   r‚   s   @r   r„   r„   Í   s¯   ø€ € € € € ð-˜ð -°4ð -ð -ð -ð -ð -ð -ð$ -2ð	/ð /à”<ð/ð ”ð/ð # T™kð	/ð
 
ˆuŒ|˜U 5¤<Ô0°4Ñ7Ð7Ô	8ð/ð /ð /ð /ð /ð /ð /ð /r   r„   c                   óž   ‡ — e Zd Zˆ fd„Z	 	 	 	 ddej        dej        dz  dej        dz  dej        dz  dedz  d	eej                 fd
„Z	ˆ xZ
S )ÚSuperGlueSelfAttentionc                 ó|  •— t          ¦   «                              ¦   «          |j        |j        z  dk    r0t	          |d¦  «        s t          d|j        › d|j        › d�¦  «        ‚|j        | _        t          |j        |j        z  ¦  «        | _        | j        | j        z  | _        t          j
        |j        | j        ¦  «        | _        t          j
        |j        | j        ¦  «        | _        t          j
        |j        | j        ¦  «        | _        t          j        |j        ¦  «        | _        |j        | _        d S )Nr   Úembedding_sizezThe hidden size (z6) is not a multiple of the number of attention heads (ú))rp   rq   r�   Únum_attention_headsÚhasattrÚ
ValueErrorr   Úattention_head_sizeÚall_head_sizer   rr   ÚqueryÚkeyÚvalueÚDropoutÚattention_probs_dropout_probÚdropoutÚ
is_decoder©rx   rl   ry   s     €r   rq   zSuperGlueSelfAttention.__init__í   s#  ø€ Ý‰Œ×ÒÑÔÐØÔ Ô :Ñ:¸aÒ?Ð?ÍÐPVÐXhÑHiÔHiÐ?Ýð8 FÔ$6ð 8ð 8Ø Ô4ð8ð 8ð 8ñô ð ð
 $*Ô#=ˆÔ Ý#& vÔ'9¸FÔ<VÑ'VÑ#WÔ#WˆÔ Ø!Ô5¸Ô8PÑPˆÔå”Y˜vÔ1°4Ô3EÑFÔFˆŒ
Ý”9˜VÔ/°Ô1CÑDÔDˆŒÝ”Y˜vÔ1°4Ô3EÑFÔFˆŒ
å”z &Ô"EÑFÔFˆŒà Ô+ˆŒˆˆr   NFr`   Úattention_maskÚencoder_hidden_statesÚencoder_attention_maskÚoutput_attentionsr   c                 ó  — |d u}|r|n|}|r|n|}|j         d         }|                      |¦  «                             |d| j        | j        ¦  «                             dd¦  «        }	|                      |¦  «                             |d| j        | j        ¦  «                             dd¦  «        }
|                      |¦  «                             |d| j        | j        ¦  «                             dd¦  «        }t          j	        ||	                     dd¦  «        ¦  «        }|t          j        | j        ¦  «        z  }|�||z   }t          j                             |d¬¦  «        }|                      |¦  «        }t          j	        ||
¦  «        }|                     dddd¦  «                             ¦   «         }|                     ¦   «         d d…         | j        fz   }|                     |¦  «        }|r||fn|f}| j        r|dz   }|S )	Nr   rB   r&   r   r|   r4   r   r   )rC   r¥   ÚviewrŸ   r¢   r}   r¦   r¤   r   ÚmatmulÚmathÚsqrtr   Ú
functionalÚsoftmaxr©   ÚpermuteÚ
contiguousr+   r£   rª   )rx   r`   r¬   r­   r®   r¯   Úis_cross_attentionÚcurrent_statesrH   Ú	key_layerÚvalue_layerÚquery_layerÚattention_scoresÚattention_probsÚcontext_layerÚnew_context_layer_shapeÚoutputss                    r   r~   zSuperGlueSelfAttention.forward  s  € ð 3¸$Ð>ÐØ2DÐWÐ.Ð.È-ˆØ3EÐYÐ/Ð/È>ˆà"Ô(¨Ô+ˆ
à�HŠH�^Ñ$Ô$ßŠT�*˜b $Ô":¸DÔ<TÑUÔUßŠY�q˜!‰_Œ_ð 	ð �JŠJ�~Ñ&Ô&ßŠT�*˜b $Ô":¸DÔ<TÑUÔUßŠY�q˜!‰_Œ_ð 	ð �JŠJ�}Ñ%Ô%ßŠT�*˜b $Ô":¸DÔ<TÑUÔUßŠY�q˜!‰_Œ_ð 	õ !œ<¨°Y×5HÒ5HÈÈRÑ5PÔ5PÑQÔQÐà+­d¬i¸Ô8PÑ.QÔ.QÑQÐØÐ%à/°.Ñ@Ðõ œ-×/Ò/Ð0@ÀbÐ/ÑIÔIˆð Ÿ,š, Ñ7Ô7ˆåœ _°kÑBÔBˆà%×-Ò-¨a°°A°qÑ9Ô9×DÒDÑFÔFˆØ"/×"4Ò"4Ñ"6Ô"6°s¸°sÔ";¸tÔ?QÐ>SÑ"SÐØ%×*Ò*Ð+BÑCÔCˆà6GÐ]�= /Ð2Ð2ÈmÐM]ˆàŒ?ð 	(Ø Ñ'ˆGØˆr   ©NNNF©rb   rc   rd   rq   r   r€   rf   r™   r   r~   r�   r‚   s   @r   r›   r›   ì   s¾   ø€ € € € € ð,ð ,ð ,ð ,ð ,ð. 48Ø:>Ø;?Ø).ð9ð 9à”|ð9ð Ô)¨DÑ0ð9ð  %Ô0°4Ñ7ð	9ð
 !&Ô 1°DÑ 8ð9ð   $™;ð9ð 
ˆuŒ|Ô	ð9ð 9ð 9ð 9ð 9ð 9ð 9ð 9r   r›   c                   óH   ‡ — e Zd Zdefˆ fd„Zdej        dej        fd„Zˆ xZS )ÚSuperGlueSelfOutputrl   c                 ó�   •— t          ¦   «                              ¦   «          t          j        |j        |j        ¦  «        | _        d S r   )rp   rq   r   rr   r�   Údenser«   s     €r   rq   zSuperGlueSelfOutput.__init__>  s6   ø€ Ý‰Œ×ÒÑÔÐÝ”Y˜vÔ1°6Ô3EÑFÔFˆŒ
ˆ
ˆ
r   r`   r   c                 ó0   — |                       |¦  «        }|S r   )rÈ   )rx   r`   Úargss      r   r~   zSuperGlueSelfOutput.forwardB  s   € ØŸ
š
 =Ñ1Ô1ˆØÐr   ©	rb   rc   rd   r   rq   r   r€   r~   r�   r‚   s   @r   rÆ   rÆ   =  sq   ø€ € € € € ðG˜ð Gð Gð Gð Gð Gð Gð U¤\ð ¸U¼\ð ð ð ð ð ð ð ð r   rÆ   Úeagerc                   óž   ‡ — e Zd Zˆ fd„Z	 	 	 	 ddej        dej        dz  dej        dz  dej        dz  dedz  d	eej                 fd
„Z	ˆ xZ
S )ÚSuperGlueAttentionc                 ó®   •— t          ¦   «                              ¦   «          t          |j                 |¦  «        | _        t          |¦  «        | _        d S r   )rp   rq   Ú SUPERGLUE_SELF_ATTENTION_CLASSESÚ_attn_implementationrx   rÆ   Úoutputr«   s     €r   rq   zSuperGlueAttention.__init__M  sD   ø€ Ý‰Œ×ÒÑÔÐÝ4°VÔ5PÔQÐRXÑYÔYˆŒ	Ý)¨&Ñ1Ô1ˆŒˆˆr   NFr`   r¬   r­   r®   r¯   r   c                 óŽ   — |                       |||||¬¦  «        }|                      |d         |¦  «        }|f|dd …         z   }|S )N©r¬   r­   r®   r¯   r   r&   )rx   rÒ   )	rx   r`   r¬   r­   r®   r¯   Úself_outputsÚattention_outputrÂ   s	            r   r~   zSuperGlueAttention.forwardR  s`   € ð —y’yØØ)Ø"7Ø#9Ø/ð !ñ 
ô 
ˆð  Ÿ;š; |°A¤¸ÑFÔFÐØ#Ð%¨°Q°R°RÔ(8Ñ8ˆØˆr   rÃ   rÄ   r‚   s   @r   rÎ   rÎ   L  s½   ø€ € € € € ð2ð 2ð 2ð 2ð 2ð 48Ø:>Ø6:Ø).ðð à”|ðð Ô)¨DÑ0ðð  %Ô0°4Ñ7ð	ð
 !&¤¨tÑ 3ðð   $™;ðð 
ˆuŒ|Ô	ðð ð ð ð ð ð ð r   rÎ   c                   óæ   ‡ — e Zd Zdeddfˆ fd„Z	 	 	 	 	 ddej        dej        dz  dej        dz  d	ej        dz  d
ededeej        eej                 dz  eej                 dz  f         fd„Z	ˆ xZ
S )ÚSuperGlueAttentionalPropagationrl   r   Nc                 óŒ  •‡‡— t          ¦   «                              ¦   «          ‰j        }t          ‰¦  «        | _        |dz  |dz  |gŠˆˆfd„t          dt          ‰¦  «        dz
  ¦  «        D ¦   «         }|                     t          j	        ‰d         ‰d         ¦  «        ¦  «         t          j
        |¦  «        | _        d S )Nr   c                 óP   •— g | ]"}t          ‰‰|d z
           ‰|         ¦  «        ‘Œ#S r‡   rˆ   )r   r‰   rl   Úmlp_channelss     €€r   r‹   z<SuperGlueAttentionalPropagation.__init__.<locals>.<listcomp>l  sC   ø€ ð 
ð 
ð 
àõ *¨&°,¸qÀ1¹uÔ2EÀ|ÐTUÄÑWÔWð
ð 
ð 
r   r&   r|   rB   )rp   rq   r�   rÎ   Ú	attentionr7   rŽ   r�   r   rr   r�   Úmlp)rx   rl   r�   r“   rÛ   ry   s    `  @€r   rq   z(SuperGlueAttentionalPropagation.__init__g  sÉ   øøø€ Ý‰Œ×ÒÑÔÐØÔ(ˆÝ+¨FÑ3Ô3ˆŒØ# a™¨°q©¸+ÐFˆð
ð 
ð 
ð 
ð 
å˜1�c ,Ñ/Ô/°!Ñ3Ñ4Ô4ð
ñ 
ô 
ˆð 	�Š•b”i ¨RÔ 0°,¸rÔ2BÑCÔCÑDÔDÐDÝ”= Ñ(Ô(ˆŒˆˆr   FÚdescriptorsr¬   r­   r®   r¯   r”   c                 óÚ   — |                       |||||¬¦  «        }|d         }|dd …         }	t          j        ||gd¬¦  «        }
|rdnd }| j        D ]} ||
¦  «        }
|r||
fz   }Œ|
||	fS )NrÔ   r   r&   r   r4   ri   )rÜ   r   r   rÝ   )rx   rÞ   r¬   r­   r®   r¯   r”   Úattention_outputsrÒ   rÜ   rz   r–   r—   s                r   r~   z'SuperGlueAttentionalPropagation.forwards  sµ   € ð !ŸNšNØØ)Ø"7Ø#9Ø/ð +ñ 
ô 
Ðð # 1Ô%ˆØ% a b bÔ)ˆ	å”y +¨vÐ!6¸AÐ>Ñ>Ô>ˆà"6Ð@˜B˜B¸DÐØ”Xð 	Hð 	HˆEØ ˜5 Ñ.Ô.ˆLØ#ð HØ$5¸¸Ñ$GÐ!øàÐ.°	Ð9Ð9r   )NNNFFr˜   r‚   s   @r   rØ   rØ   f  sö   ø€ € € € € ð
)˜ð 
)°4ð 
)ð 
)ð 
)ð 
)ð 
)ð 
)ð /3Ø59Ø6:Ø"'Ø%*ð:ð :à”\ð:ð œ tÑ+ð:ð  %œ|¨dÑ2ð	:ð
 !&¤¨tÑ 3ð:ð  ð:ð #ð:ð 
ˆuŒ|˜U 5¤<Ô0°4Ñ7¸¸u¼|Ô9LÈtÑ9SÐSÔ	Tð:ð :ð :ð :ð :ð :ð :ð :r   rØ   c                   ó”   ‡ — e Zd Zdeddfˆ fd„Z	 	 	 ddej        dej        dz  ded	edz  deej        edz  edz  f         f
d
„Z	ˆ xZ
S )ÚSuperGlueAttentionalGNNrl   r   Nc                 ó  •‡— t          ¦   «                              ¦   «          ‰j        | _        ‰j        | _        t          j        ˆfd„t          t          | j        ¦  «        ¦  «        D ¦   «         ¦  «        | _	        d S )Nc                 ó.   •— g | ]}t          ‰¦  «        ‘ŒS ri   )rØ   )r   r<   rl   s     €r   r‹   z4SuperGlueAttentionalGNN.__init__.<locals>.<listcomp>–  s#   ø€ Ð$tÐ$tÐ$tÐQRÕ%DÀVÑ%LÔ%LÐ$tÐ$tÐ$tr   )
rp   rq   r�   Úgnn_layers_typesÚlayers_typesr   r�   r7   rŽ   r“   r«   s    `€r   rq   z SuperGlueAttentionalGNN.__init__’  sp   øø€ Ý‰Œ×ÒÑÔÐØ!Ô-ˆÔØ"Ô3ˆÔÝ”mÐ$tÐ$tÐ$tÐ$tÕV[Õ\_Ð`dÔ`qÑ\rÔ\rÑVsÔVsÐ$tÑ$tÔ$tÑuÔuˆŒˆˆr   FrÞ   r_   r¯   r”   c           	      ó.  — |rdnd }|rdnd }|j         \  }}}	|r||fz   }t          | j        | j        ¦  «        D ]×\  }
}d }d }|dk    r�|                     dd|| j        ¦  «                             d¦  «                             ||| j        ¦  «        }|�B|                     dddd|¦  «                             d¦  «                             |dd|¦  «        nd } |
||||||¬¦  «        }|d         }|r||d         z   }|r||d         z   }||z   }ŒØ|||fS )Nri   ÚcrossrB   r   r&   )r¬   r­   r®   r”   r¯   r   )rC   r   r“   ræ   Úreshaper�   Úflip)rx   rÞ   r_   r¯   r”   r–   Úall_attentionsrH   Únum_keypointsr<   Ú	gnn_layerÚ
layer_typer­   r®   Úgnn_outputsÚdeltas                   r   r~   zSuperGlueAttentionalGNN.forward˜  s�  € ð #7Ð@˜B˜B¸DÐØ0Ð:˜˜°dˆà'2Ô'8Ñ$ˆ
�M 1Øð 	CØ 1°[°NÑ BÐå%(¨¬°dÔ6GÑ%HÔ%Hð 	.ð 	.Ñ!ˆI�zØ$(Ð!Ø%)Ð"Ø˜WÒ$Ð$à×'Ò'¨¨A¨}¸dÔ>NÑOÔOß’T˜!‘W”Wß’W˜Z¨¸Ô8HÑIÔIð &ð Ð'ð —L’L  Q¨¨1¨mÑ<Ô<×AÒAÀ!ÑDÔD×LÒLÈZÐYZÐ\]Ð_lÑmÔmÐmàð 'ð $˜)ØØ#Ø&;Ø'=Ø%9Ø"3ðñ ô ˆKð   ”NˆEà#ð GØ$5¸ÀA¼Ñ$FÐ!Ø ð AØ!/°+¸a´.Ñ!@�à%¨Ñ-ˆKˆKØÐ-¨~Ð=Ð=r   )NFFr˜   r‚   s   @r   râ   râ   ‘  sÊ   ø€ € € € € ðv˜ð v°4ð vð vð vð vð vð vð %)Ø"'Ø,1ð->ð ->à”\ð->ð Œl˜TÑ!ð->ð  ð	->ð
 # T™kð->ð 
ˆuŒ|˜U T™\¨5°4©<Ð7Ô	8ð->ð ->ð ->ð ->ð ->ð ->ð ->ð ->r   râ   c                   óL   ‡ — e Zd Zdeddfˆ fd„Zdej        dej        fd„Zˆ xZS )ÚSuperGlueFinalProjectionrl   r   Nc                 óŽ   •— t          ¦   «                              ¦   «          |j        }t          j        ||d¬¦  «        | _        d S )NT)Úbias)rp   rq   r�   r   rr   Ú
final_proj)rx   rl   r�   ry   s      €r   rq   z!SuperGlueFinalProjection.__init__É  s=   ø€ Ý‰Œ×ÒÑÔÐØÔ(ˆÝœ) K°À4ÐHÑHÔHˆŒˆˆr   rÞ   c                 ó,   — |                       |¦  «        S r   )rõ   )rx   rÞ   s     r   r~   z SuperGlueFinalProjection.forwardÎ  s   € Ø�Š˜{Ñ+Ô+Ð+r   rË   r‚   s   @r   rò   rò   È  sy   ø€ € € € € ðI˜ð I°4ð Ið Ið Ið Ið Ið Ið
, 5¤<ð ,°E´Lð ,ð ,ð ,ð ,ð ,ð ,ð ,ð ,r   rò   c                   ót   ‡ — e Zd ZU eed<   dZdZdZ ej	        ¦   «         de
j        ddfˆ fd„¦   «         Zˆ xZS )	ÚSuperGluePreTrainedModelrl   Ú	superglueÚpixel_values)ÚimageÚmoduler   Nc                 ó    •— t          ¦   «                              |¦  «         t          |d¦  «        rt          j        |j        ¦  «         dS dS )zInitialize the weightsÚ	bin_scoreN)rp   Ú_init_weightsr    ÚinitÚones_rþ   )rx   rü   ry   s     €r   rÿ   z&SuperGluePreTrainedModel._init_weightsÙ  sR   ø€ õ 	‰Œ×Ò˜fÑ%Ô%Ð%Ý�6˜;Ñ'Ô'ð 	)ÝŒJ�vÔ'Ñ(Ô(Ð(Ð(Ð(ð	)ð 	)r   )rb   rc   rd   r   rg   Úbase_model_prefixÚmain_input_nameÚinput_modalitiesr   Úno_gradr   ÚModulerÿ   r�   r‚   s   @r   rø   rø   Ò  s~   ø€ € € € € € àÐÐÑØ#ÐØ$€OØ!Ðà€U„]�_„_ð) B¤Ið )°$ð )ð )ð )ð )ð )ñ „_ð)ð )ð )ð )ð )r   rø   zV
    SuperGlue model taking images as inputs and outputting the matching of them.
    c                   ó.  ‡ — e Zd ZdZdeddfˆ fd„Z	 	 	 ddej        dej        dej        d	ed
edej        dz  de	dz  de	dz  de
ej        ej        e
e
f         fd„Ze	 	 	 	 ddej        dej        dz  de	dz  de	dz  de	dz  de
ez  fd„¦   «         Zˆ xZS )ÚSuperGlueForKeypointMatchingaø  SuperGlue feature matching middle-end

    Given two sets of keypoints and locations, we determine the
    correspondences by:
      1. Keypoint Encoding (normalization + visual feature and location fusion)
      2. Graph Neural Network with multiple self and cross-attention layers
      3. Final projection layer
      4. Optimal Transport Layer (a differentiable Hungarian matching algorithm)
      5. Thresholding matrix based on mutual exclusivity and a match_threshold

    The correspondence ids use -1 to indicate non-matching points.

    Paul-Edouard Sarlin, Daniel DeTone, Tomasz Malisiewicz, and Andrew
    Rabinovich. SuperGlue: Learning Feature Matching with Graph Neural
    Networks. In CVPR, 2020. https://huggingface.co/papers/1911.11763
    rl   r   Nc                 ó´  •— t          ¦   «                              |¦  «         t          j        |j        ¦  «        | _        t          |¦  «        | _        t          |¦  «        | _	        t          |¦  «        | _        t          j                             t          j        d¦  «        ¦  «        }|                      d|¦  «         |                      ¦   «          d S )Ng      ð?rþ   )rp   rq   r   Úfrom_configÚkeypoint_detector_configÚkeypoint_detectorr„   Úkeypoint_encoderrâ   Úgnnrò   Úfinal_projectionr   r   Ú	Parameterr(   Úregister_parameterÚ	post_init)rx   rl   rþ   ry   s      €r   rq   z%SuperGlueForKeypointMatching.__init__ø  s«   ø€ Ý‰Œ×Ò˜Ñ Ô Ð å!>Ô!JÈ6ÔKjÑ!kÔ!kˆÔå 8¸Ñ @Ô @ˆÔÝ*¨6Ñ2Ô2ˆŒÝ 8¸Ñ @Ô @ˆÔå”H×&Ò&¥u¤|°CÑ'8Ô'8Ñ9Ô9ˆ	Ø×Ò ¨YÑ7Ô7Ð7à�ŠÑÔÐÐÐr   r    rÞ   r>   r!   r"   r_   r¯   r”   c	                 ó6
  ‡&‡'— |rdnd}	|rdnd}
|j         d         dk    rH|j         dd…         }|                     |dt          j        ¬¦  «        |                     |¦  «        |	|
fS |j         \  Š&}Š'}|                     ‰&dz  ‰'d¦  «        }|                     ‰&dz  ‰'| j        j        ¦  «        }|                     ‰&dz  ‰'¦  «        }|�|                     ‰&dz  ‰'¦  «        nd}t          |||¦  «        }|  	                    |||¬¦  «        }|d         }||z   }t          | j        |dd…dd…dd…f         |¬	¦  «        }|                      ||||¬
¦  «        }|d         }|                      |¦  «        }|                     ‰&d‰'| j        j        ¦  «        }|dd…df         }|dd…df         }||                     dd¦  «        z  }|| j        j        dz  z  }|� |                     ‰&d‰'¦  «        }|dd…df                              d¦  «        }|dd…df                              d¦  «        }t          j        ||¦  «        }|                     |dk    t          j        |j        ¦  «        j        ¦  «        }t)          || j        | j        j        ¬¦  «        }|dd…dd…dd…f                              d¦  «        }|dd…dd…dd…f                              d¦  «        }|j        }|j        }t3          |d¦  «        d         |                     d|¦  «        k    }t3          |d¦  «        d         |                     d|¦  «        k    }|                     d¦  «        }t          j        ||j                             ¦   «         |¦  «        }t          j        || j        j        k    ||¦  «        }t          j        ||                     d|¦  «        |¦  «        }|||k    z  } ||                      d|¦  «        z  }!t          j        | ||                     d¦  «        ¦  «        }"t          j        |!||                     d¦  «        ¦  «        }#t          j         |"|#gd¬¦  «                             ‰&dd¦  «        }$t          j         ||gd¬¦  «                             ‰&dd¦  «        }%|r8|	|d         z   }	|	|d         z   }	|	|fz   }	tC          ˆ&ˆ'fd„|	D ¦   «         ¦  «        }	|r'|
|d         z   }
tC          ˆ&ˆ'fd„|
D ¦   «         ¦  «        }
|$|%|	|
fS )a=  
        Perform keypoint matching between two images.

        Args:
            keypoints (`torch.Tensor` of shape `(batch_size, 2, num_keypoints, 2)`):
                Keypoints detected in the pair of image.
            descriptors (`torch.Tensor` of shape `(batch_size, 2, descriptor_dim, num_keypoints)`):
                Descriptors of the keypoints detected in the image pair.
            scores (`torch.Tensor` of shape `(batch_size, 2, num_keypoints)`):
                Confidence scores of the keypoints detected in the image pair.
            height (`int`): Image height.
            width (`int`): Image width.
            mask (`torch.Tensor` of shape `(batch_size, 2, num_keypoints)`, *optional*):
                Mask indicating which values in the keypoints, matches and matching_scores tensors are keypoint matching
                information.
            output_attentions (`bool`, *optional*):
                Whether or not to return the attentions tensors. Default to `config.output_attentions`.
            output_hidden_states (`bool`, *optional*):
                Whether or not to return the hidden states of all layers. Default to `config.output_hidden_states`.

        Returns:
            matches (`torch.Tensor` of shape `(batch_size, 2, num_keypoints)`):
                For each image pair, for each keypoint in image0, the index of the keypoint in image1 that was matched
                with. And for each keypoint in image1, the index of the keypoint in image0 that was matched with.
            matching_scores (`torch.Tensor` of shape `(batch_size, 2, num_keypoints)`):
                Scores of predicted matches for each image pair
            all_hidden_states (`tuple(torch.FloatTensor)`, *optional*):
                Tuple of `torch.FloatTensor` (one for the output of each stage) of shape `(1, 2, num_keypoints,
                num_channels)`.
            all_attentions (`tuple(torch.FloatTensor)`, *optional*):
                Tuple of `torch.FloatTensor` (one for each layer) of shape `(1, 2, num_heads, num_keypoints,
                num_keypoints)`.
        ri   Nr   r   rB   )r%   )r”   r&   )rl   Úinputs_embedsr¬   )r_   r”   r¯   g      à?)r@   r4   c              3   óp   •K  — | ]0}|                      ‰d ‰d¦  «                             dd¦  «        V — Œ1dS )r   rB   r|   N)ré   r}   ©r   rW   rH   rì   s     €€r   r   zASuperGlueForKeypointMatching._match_image_pair.<locals>.<genexpr>‹  sV   øè è € ð &ð &ØRS�—	’	˜* a¨¸Ñ;Ô;×EÒEÀbÈ"ÑMÔMð&ð &ð &ð &ð &ð &r   c              3   óJ   •K  — | ]}|                      ‰d d‰‰¦  «        V — ŒdS )r   rB   N)ré   r  s     €€r   r   zASuperGlueForKeypointMatching._match_image_pair.<locals>.<genexpr>�  s8   øè è € Ð"vÐ"vÐbc 1§9¢9¨Z¸¸BÀÈ}Ñ#]Ô#]Ð"vÐ"vÐ"vÐ"vÐ"vÐ"vr   )"rC   Únew_fullr   r   Ú	new_zerosré   rl   r�   r.   r  r	   r  r  r}   r9   Úlogical_andÚmasked_fillÚfinfor%   ÚminrS   rþ   Úsinkhorn_iterationsr)   ÚindicesrX   ÚgatherrD   Úwherer*   ÚexpÚmatching_thresholdr   r   )(rx   r    rÞ   r>   r!   r"   r_   r¯   r”   r–   rë   rC   r<   Úencoded_keypointsÚlast_hidden_stateÚextended_attention_maskrï   Úprojected_descriptorsÚfinal_descriptorsÚfinal_descriptors0Úfinal_descriptors1Úmask0Úmask1Úmax0Úmax1Úindices0Úindices1Úmutual0Úmutual1ÚzeroÚmatching_scores0Úmatching_scores1Úvalid0Úvalid1Úmatches0Úmatches1r]   r^   rH   rì   s(                                         @@r   Ú_match_image_pairz.SuperGlueForKeypointMatching._match_image_pair  s±  øø€ ðX #7Ð@˜B˜B¸DÐØ0Ð:˜˜°dˆàŒ?˜1Ô Ò"Ð"Ø”O C R CÔ(ˆEà×"Ò" 5¨"µE´IÐ"Ñ>Ô>Ø×#Ò# EÑ*Ô*Ø!Øð	ð ð +4¬/Ñ'ˆ
�A�} aà×%Ò% j°1¡n°mÀQÑGÔGˆ	Ø!×)Ò)¨*°q©.¸-ÈÌÔI`ÑaÔaˆØ—’ 
¨Q¡°Ñ>Ô>ˆØ>BÐ>Nˆt�|Š|˜J¨™N¨MÑ:Ô:Ð:ÐTXˆõ (¨	°6¸5ÑAÔAˆ	à ×1Ò1°)¸VÐZnÐ1ÑoÔoÐà-¨aÔ0Ðð "Ð$5Ñ5ˆå";Ø”;Ø% a a a¨¨1¨¨a¨a¨a iÔ0Øð#
ñ #
ô #
Ðð —h’hØØ(Ø!5Ø/ð	 ñ 
ô 
ˆð " !”nˆð !%× 5Ò 5°kÑ BÔ BÐð 2×9Ò9¸*ÀaÈÐX\ÔXcÔXoÑpÔpÐØ.¨q¨q¨q°!¨tÔ4ÐØ.¨q¨q¨q°!¨tÔ4Ðð $Ð&8×&BÒ&BÀ1ÀaÑ&HÔ&HÑHˆØ˜$œ+Ô1°3Ñ6Ñ6ˆàÐØ—<’< 
¨A¨}Ñ=Ô=ˆDØ˜˜˜˜A˜”J×(Ò(¨Ñ+Ô+ˆEØ˜˜˜˜A˜”J×(Ò(¨Ñ+Ô+ˆEÝÔ$ U¨EÑ2Ô2ˆDØ×'Ò'¨°ª	µ5´;¸v¼|Ñ3LÔ3LÔ3PÑQÔQˆFõ ' v¨t¬~È$Ì+ÔJiÐjÑjÔjˆð �a�a�a˜˜"˜˜c˜r˜c�kÔ"×&Ò& qÑ)Ô)ˆØ�a�a�a˜˜"˜˜c˜r˜c�kÔ"×&Ò& qÑ)Ô)ˆØ”<ˆØ”<ˆÝ˜h¨Ñ*Ô*¨4Ô0°H·O²OÀAÀxÑ4PÔ4PÒPˆÝ˜h¨Ñ*Ô*¨4Ô0°H·O²OÀAÀxÑ4PÔ4PÒPˆØ× Ò  Ñ#Ô#ˆÝ œ; w°´·²Ñ0AÔ0AÀ4ÑHÔHÐÝ œ;Ð'7¸$¼+Ô:XÒ'XÐZjÐlpÑqÔqÐÝ œ; wÐ0@×0GÒ0GÈÈ8Ñ0TÔ0TÐVZÑ[Ô[ÐØÐ,¨tÒ3Ñ4ˆØ˜6Ÿ=š=¨¨HÑ5Ô5Ñ5ˆÝ”;˜v x°×1DÒ1DÀRÑ1HÔ1HÑIÔIˆÝ”;˜v x°×1DÒ1DÀRÑ1HÔ1HÑIÔIˆå”)˜X xÐ0°aÐ8Ñ8Ô8×@Ò@ÀÈQÐPRÑSÔSˆÝœ)Ð%5Ð7GÐ$HÈaÐPÑPÔP×XÒXÐYcÐefÐhjÑkÔkˆàð 	Ø 1Ð4EÀaÔ4HÑ HÐØ 1°KÀ´NÑ BÐØ 1Ð5JÐ4LÑ LÐÝ %ð &ð &ð &ð &ð &ØWhð&ñ &ô &ñ !ô !Ðð ð 	wØ+¨k¸!¬nÑ<ˆNÝ"Ð"vÐ"vÐ"vÐ"vÐ"vÐguÐ"vÑ"vÔ"vÑvÔvˆNð ØØØð	
ð 	
r   rú   ÚlabelsÚreturn_dictc           
      ó,  — d}|�t          d¦  «        ‚|�|n| j        j        }|�|n| j        j        }|�|n| j        j        }|j        dk    s|                     d¦  «        dk    rt          d¦  «        ‚|j        \  }}	}
}}|                     |dz  |
||¦  «        }|  	                    |¦  «        }|dd…         \  }}}}|                     |ddd¦  «         
                    |¦  «        }|                     |dd¦  «         
                    |¦  «        }|                     |dd| j        j        ¦  «         
                    |¦  «        }|                     |dd¦  «        }|                     ¦   «         }|dd…dd…dd…d	f         |z  |dd…dd…dd…d	f<   |dd…dd…dd…df         |z  |dd…dd…dd…df<   |                      ||||||||¬
¦  «        \  }}}}|s t          d„ |||||||fD ¦   «         ¦  «        S t          |||||||¬¦  «        S )aŸ  
        Examples:

        ```python
        >>> from transformers import AutoImageProcessor, AutoModel
        >>> import torch
        >>> from PIL import Image
        >>> import httpx
        >>> from io import BytesIO

        >>> url = "https://github.com/magicleap/SuperGluePretrainedNetwork/blob/master/assets/phototourism_sample_images/london_bridge_78916675_4568141288.jpg?raw=true"
        >>> with httpx.stream("GET", url) as response:
        ...     image_1 = Image.open(BytesIO(response.read()))

        >>> url = "https://github.com/magicleap/SuperGluePretrainedNetwork/blob/master/assets/phototourism_sample_images/london_bridge_19481797_2295892421.jpg?raw=true"
        >>> with httpx.stream("GET", url) as response:
        ...     image_2 = Image.open(BytesIO(response.read()))

        >>> images = [image_1, image_2]

        >>> processor = AutoImageProcessor.from_pretrained("magic-leap-community/superglue_outdoor")
        >>> model = AutoModel.from_pretrained("magic-leap-community/superglue_outdoor")

        >>> with torch.no_grad():
        >>>     inputs = processor(images, return_tensors="pt")
        >>>     outputs = model(**inputs)
        ```Nz9SuperGlue is not trainable, no labels should be provided.é   r&   r   zOInput must be a 5D tensor of shape (batch_size, 2, num_channels, height, width)é   rB   r   )r_   r¯   r”   c              3   ó   K  — | ]}|®|V — Œ	d S r   ri   )r   Úvs     r   r   z7SuperGlueForKeypointMatching.forward.<locals>.<genexpr>å  s0   è è € ð ð àØ�=ð à �=�=�=ðð r   )r\   r]   r^   r    r_   r`   ra   )r¡   rl   r¯   r”   r<  Úndimr+   rC   ré   r  rE   r�   Úcloner:  r   r[   )rx   rú   r;  r¯   r”   r<  Úkwargsr\   rH   r<   Úchannelsr!   r"   Úkeypoint_detectionsr    r>   rÞ   r_   Úabsolute_keypointsr]   r^   r`   ra   s                          r   r~   z$SuperGlueForKeypointMatching.forward™  s¼  € ðJ ˆØÐÝÐXÑYÔYÐYà1BÐ1NÐ-Ð-ÐTXÔT_ÔTqÐà$8Ð$DÐ Ð È$Ì+ÔJjð 	ð &1Ð%<�k�kÀ$Ä+ÔBYˆàÔ Ò!Ð! \×%6Ò%6°qÑ%9Ô%9¸QÒ%>Ð%>ÝÐnÑoÔoÐoà1=Ô1CÑ.ˆ
�A�x ¨Ø#×+Ò+¨J¸©N¸HÀfÈeÑTÔTˆØ"×4Ò4°\ÑBÔBÐà/BÀ2ÀAÀ2Ô/FÑ,ˆ	�6˜;¨Ø×%Ò% j°!°R¸Ñ;Ô;×>Ò>¸|ÑLÔLˆ	Ø—’ 
¨A¨rÑ2Ô2×5Ò5°lÑCÔCˆØ!×)Ò)¨*°a¸¸T¼[Ô=TÑUÔU×XÒXÐYeÑfÔfˆØ�|Š|˜J¨¨2Ñ.Ô.ˆà&Ÿ_š_Ñ.Ô.ÐØ);¸A¸A¸A¸q¸q¸qÀ!À!À!ÀQ¸JÔ)GÈ%Ñ)OÐ˜1˜1˜1˜a˜a˜a    A˜:Ñ&Ø);¸A¸A¸A¸q¸q¸qÀ!À!À!ÀQ¸JÔ)GÈ&Ñ)PÐ˜1˜1˜1˜a˜a˜a    A˜:Ñ&à>B×>TÒ>TØØØØØØØ/Ø!5ð ?Uñ 	?
ô 	?
Ñ;ˆ� -°ð ð 	Ýð ð à ¨¸)ÀTÈ=ÐZdÐeðñ ô ñ ô ð õ /ØØØ+ØØØ'Ø!ð
ñ 
ô 
ð 	
r   )NNN)NNNN)rb   rc   rd   re   r   rq   r   r€   r   r™   r   r:  r   rf   Ú
LongTensorr[   r~   r�   r‚   s   @r   r  r  á  s   ø€ € € € € ðð ð"˜ð °4ð ð ð ð ð ð ð* %)Ø)-Ø,0ðQ
ð Q
à”<ðQ
ð ”\ðQ
ð ”ð	Q
ð
 ðQ
ð ðQ
ð Œl˜TÑ!ðQ
ð   $™;ðQ
ð # T™kðQ
ð 
ˆuŒ|˜Uœ\¨5°%Ð7Ô	8ðQ
ð Q
ð Q
ð Q
ðf ð +/Ø)-Ø,0Ø#'ðY
ð Y
àÔ'ðY
ð Ô  4Ñ'ðY
ð   $™;ð	Y
ð
 # T™kðY
ð ˜D‘[ðY
ð 
Ð0Ñ	0ðY
ð Y
ð Y
ñ „^ðY
ð Y
ð Y
ð Y
ð Y
r   r  ).re   r³   Údataclassesr   r   r   Útransformersr   Ú5transformers.models.superglue.configuration_supergluer   Ú r   r   Úmasking_utilsr	   Úutilsr
   r   r   Úautor   Ú
get_loggerrb   Úloggerr   r€   r   r   r.   r=   rS   rX   r[   r  rk   r„   r›   rÆ   rÐ   rÎ   rØ   râ   rò   rø   r  Ú__all__ri   r   r   ú<module>rS     sz  ðð Ð à €€€Ø !Ð !Ð !Ð !Ð !Ð !à €€€Ø Ð Ð Ð Ð Ð à (Ð (Ð (Ð (Ð (Ð (Ø QÐ QÐ QÐ QÐ QÐ Qà &Ð &Ð &Ð &Ð &Ð &Ø 6Ð 6Ð 6Ð 6Ð 6Ð 6Ø 9Ð 9Ð 9Ð 9Ð 9Ð 9Ð 9Ð 9Ð 9Ð 9Ø 0Ð 0Ð 0Ð 0Ð 0Ð 0ð 
ˆÔ	˜HÑ	%Ô	%€ðj  e¤lÔ 3ð jÀEÈ%Ì,ÔDWð jÐ\aÐbgÔbnÔ\oð jð jð jð jð B 5¤<ð B¸ð BÀSð BÈUÌ\ð Bð Bð Bð Bð*UØ”\ðUà"œ\ðUð #œ\ðUð ð	Uð
 „\ðUð Uð Uð Uð:*( %¤,ð *(¸5¼<ð *(ÐUXð *(Ð]bÔ]ið *(ð *(ð *(ð *(ðZ2˜ð 2 ¤ð 2ð 2ð 2ð 2ð €ððñ ô ð ð7ð 7ð 7ð 7ð 7 kñ 7ô 7ñ „ñô ð7ð<ð ð ð ð  B¤Iñ ô ð ð /ð /ð /ð /ð /˜rœyñ /ô /ð /ð>Nð Nð Nð Nð N˜RœYñ Nô Nð Nðbð ð ð ð ˜"œ)ñ ô ð ð Ð#ð$Ð  ð
ð ð ð ð ˜œñ ô ð ð4(:ð (:ð (:ð (:ð (: b¤iñ (:ô (:ð (:ðV4>ð 4>ð 4>ð 4>ð 4>˜bœiñ 4>ô 4>ð 4>ðn,ð ,ð ,ð ,ð ,˜rœyñ ,ô ,ð ,ð ð)ð )ð )ð )ð )˜ñ )ô )ñ „ð)ð €ððñ ô ð
M
ð M
ð M
ð M
ð M
Ð#;ñ M
ô M
ñô ð
M
ð` &Ð'EÐ
F€€€r   