§
    9²¸jë  ã                   ó,  — d 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
mZmZ ddlmZmZ dd	lmZmZ  G d
„ ded¬¦  «        Z	 d dedededededeeef         fd„Zdddeddfd„Z	 d!dddededed         fd„Ze G d„ de¦  «        ¦   «         ZdgZdS )"z"Image processor class for SigLIP2.é    N)Ú
functionalé   )ÚTorchvisionBackend)ÚBatchFeature)Ú
ImageInputÚPILImageResamplingÚSizeDict)ÚImagesKwargsÚUnpack)Ú
TensorTypeÚauto_docstringc                   ó(   — e Zd ZU dZeed<   eed<   dS )ÚSiglip2ImageProcessorKwargsar  
    patch_size (`int`, *optional*, defaults to `self.patch_size`):
        The size (resolution) of each patch the image will be split to.
    max_num_patches (`int`, *optional*, defaults to `self.max_num_patches`):
        The image will be resized to have at most this number of patches,
        and then padded in "patch" dimension to match this number exactly.
    Ú
patch_sizeÚmax_num_patchesN)Ú__name__Ú
__module__Ú__qualname__Ú__doc__ÚintÚ__annotations__© ó    ú„/root/private/project/pallet_detection_v2/.venv/lib/python3.11/site-packages/transformers/models/siglip2/image_processing_siglip2.pyr   r      s3   € € € € € € ðð ð €O€O�OØÐÐÑÐÐr   r   F)Útotalçñhãˆµøä>Úimage_heightÚimage_widthr   r   ÚepsÚreturnc                 ó2  ‡— ddl Šdt          dt          dt          dt          fˆfd„}|dz  d	}}||z
  |k    rA||z   d
z  } ||| |¦  «        }	 ||||¦  «        }
|	|z  |
|z  z  }||k    r|}n|}||z
  |k    °A|} ||| |¦  «        }	 ||||¦  «        }
|	|
fS )a"  
    Determine image size based on max number of patches, ensure dimensions are divisible by patch size and image is at least 1 patch.

    Args:
        image_height (`int`):
            Original image height.
        image_width (`int`):
            Original image width.
        patch_size (`int`):
            Patch size for processing.
        max_num_patches (`int`):
            Maximum number of patches.
        eps (`float`):
            Small threshold for binary search.

    Returns:
        Tuple: (target_height, target_width)
    r   NÚscaleÚsizer   r    c                 ó‚   •— || z  }‰                      ||z  ¦  «        |z  }t          ||¦  «        }t          |¦  «        S ©N)ÚceilÚmaxr   )r"   r#   r   Úscaled_sizeÚmaths       €r   Úget_scaled_image_sizezAget_image_size_for_max_num_patches.<locals>.get_scaled_image_sizeA   sD   ø€ Ø˜U‘lˆØ—i’i ¨jÑ 8Ñ9Ô9¸JÑFˆÝ˜* kÑ2Ô2ˆÝ�;ÑÔÐr   é
   g      Y@é   )r)   Úfloatr   )r   r   r   r   r   r*   Ú	scale_minÚ	scale_maxr"   Útarget_heightÚtarget_widthÚnum_patchesr)   s               @r   Ú"get_image_size_for_max_num_patchesr3   *   s  ø€ ð* €K€K€Kð ¥Uð  µ#ð  Å3ð  Í3ð  ð  ð  ð  ð  ð  ð  ™8 Uˆy€IØ�yÑ  SÒ
(Ð
(Ø˜YÑ&¨!Ñ+ˆØ-Ð-¨e°\À:ÑNÔNˆØ,Ð,¨U°KÀÑLÔLˆØ$ zÑ1°lÀZÑ6OÑPˆà˜/Ò)Ð)ØˆIˆIàˆIð �yÑ  SÒ
(Ð
(ð €EØ)Ð)¨%°¸zÑJÔJ€MØ(Ð(¨°¸ZÑHÔH€LØ˜,Ð&Ð&r   Úimageútorch.Tensorc                 óÆ   — | j         \  }}}||z  }||z  }|                      |||||¦  «        }|                     ddddd¦  «        }|                     ||z  d¦  «        }|S )zÍ
    Convert 3D tensor image of shape (num_channels, image_height, image_width) into 2D tensor of patches of shape
    (num_patches_height * num_patches_width, patch_size * patch_size * num_channels).
    é   r   r,   é   r   éÿÿÿÿ)ÚshapeÚreshapeÚpermute)r4   r   Únum_channelsr   r   Únum_patches_heightÚnum_patches_widthÚpatched_images           r   Úconvert_image_to_patchesrA   Z   s‚   € ð
 /4¬kÑ+€L�, Ø%¨Ñ3ÐØ# zÑ1ÐØ—M’M ,Ð0BÀJÐPaÐcmÑnÔn€MØ!×)Ò)¨!¨Q°°1°aÑ8Ô8€MØ!×)Ò)Ð*<Ð?PÑ*PÐRTÑUÔU€MØÐr   ÚtensorÚtarget_lengthÚ	pad_value)r5   r5   c                 ó  — | j         d         }||z
  }t          j        |ft          j        ¬¦  «        }|dk    rDddg| j        dz
  z  d|gz   }t          j        j                             | |d|¬¦  «        } d|| d…<   | |fS )z3
    Pad the tensor along the first dimension.
    r   )Údtyper7   Úconstant)ÚmodeÚvalueN)r:   ÚtorchÚonesÚint32ÚndimÚnnr   Úpad)rB   rC   rD   Úcurrent_lengthÚpadding_lengthÚmaskÚpaddings          r   Úpad_along_first_dimrT   h   s™   € ð ”\ !”_€NØ" ^Ñ3€NÝŒ:�}Ð&­e¬kÐ:Ñ:Ô:€DØ˜ÒÐØ�a�&˜FœK¨!™OÑ,°°>Ð/BÑBˆÝ”Ô$×(Ò(¨°¸zÐQZÐ(Ñ[Ô[ˆØ!"ˆˆnˆ_ÐÐÑØ�4ˆ<Ðr   c                   ó2  ‡ — e Zd ZeZej        Zg d¢Zg d¢Z	dZ
dZdZdZdZg d¢Zdee         fˆ fd„Zededee         d	efˆ fd
„¦   «         Zd	efˆ fd„Zded         dedededddedededeee         z  dz  deee         z  dz  deez  dz  d	efd„Zˆ xZS )ÚSiglip2ImageProcessor)ç      à?rW   rW   Té   é   ©Úpixel_valuesÚpixel_attention_maskÚspatial_shapesÚkwargsc                 ó:   •—  t          ¦   «         j        di |¤Ž d S )Nr   )ÚsuperÚ__init__©Úselfr^   Ú	__class__s     €r   ra   zSiglip2ImageProcessor.__init__…   s&   ø€ Ø�‰ŒÔÐ"Ð"˜6Ð"Ð"Ð"Ð"Ð"r   Úimagesr    c                 ó8   •—  t          ¦   «         j        |fi |¤ŽS r%   )r`   Ú
preprocess)rc   re   r^   rd   s      €r   rg   z Siglip2ImageProcessor.preprocessˆ   s"   ø€ à!�u‰wŒwÔ! &Ð3Ð3¨FÐ3Ð3Ð3r   c                 ób   •— |                      dd ¦  «          t          ¦   «         j        di |¤ŽS )NÚ	do_resizer   )Úpopr`   Ú_validate_preprocess_kwargsrb   s     €r   rk   z1Siglip2ImageProcessor._validate_preprocess_kwargsŒ   s4   ø€ à�
Š
�; Ñ%Ô%Ð%Ø2�u‰wŒwÔ2Ð<Ð<°VÐ<Ð<Ð<r   r5   ri   r   r   Úresamplez7PILImageResampling | tvF.InterpolationMode | int | NoneÚ
do_rescaleÚrescale_factorÚdo_normalizeÚ
image_meanNÚ	image_stdÚreturn_tensorsc           	      ó0  — g }g }g }|D ]÷}|rUt          |j        d         |j        d         ||¬¦  «        \  }}t          ||¬¦  «        }|                      |||¬¦  «        }|                      |||||	|
¦  «        }t          ||¦  «        }t          ||¦  «        \  }}|j        d         |z  }|j        d         |z  }|                     ||f¦  «         |                     |¦  «         |                     |¦  «         Œøt          |||dœ|¬¦  «        }|S )Néþÿÿÿr9   )r   r   r   r   )ÚheightÚwidth)r4   r#   rl   rZ   )ÚdataÚtensor_type)	r3   r:   r	   ÚresizeÚrescale_and_normalizerA   rT   Úappendr   )rc   re   ri   r   r   rl   rm   rn   ro   rp   rq   rr   r^   Úpixel_masksr[   r]   r4   ru   rv   Ú	size_dictÚpatchesrR   r>   r?   Úbatch_features                            r   Ú_preprocessz!Siglip2ImageProcessor._preprocess‘   sf  € ð ˆØˆØˆàð 	%ð 	%ˆEØð TÝ BØ!&¤¨R¤Ø %¤¨B¤Ø)Ø$3ð	!ñ !ô !‘�˜õ %¨F¸%Ð@Ñ@Ô@�	ØŸš¨%°iÈ(˜ÑSÔS�à×.Ò.¨u°jÀ.ÐR^Ð`jÐluÑvÔvˆEõ /¨u°jÑAÔAˆGÝ/°¸ÑIÔI‰MˆG�Tà!&¤¨R¤°JÑ!>ÐØ %¤¨B¤°:Ñ =Ðà×!Ò!Ð#5Ð7HÐ"IÑJÔJÐJØ×Ò Ñ(Ô(Ð(Ø×Ò˜tÑ$Ô$Ð$Ð$å$à ,Ø(3Ø"0ðð ð
 'ð
ñ 
ô 
ˆð Ðr   ) r   r   r   r   Úvalid_kwargsr   ÚBILINEARrl   rp   rq   ri   rm   ro   r   r   Úmodel_input_namesr   ra   r   r   r   rg   Útuplerk   ÚlistÚboolr   r-   Ústrr   r€   Ú__classcell__)rd   s   @r   rV   rV   x   s¯  ø€ € € € € à.€LØ!Ô*€HØ ��€JØ��€IØ€IØ€JØ€LØ€JØ€OØRÐRÐRÐð# Ð(CÔ!Dð #ð #ð #ð #ð #ð #ð ð4 ð 4°vÐ>YÔ7Zð 4Ð_kð 4ð 4ð 4ð 4ð 4ñ „^ð4ð=°uð =ð =ð =ð =ð =ð =ð
3à�^Ô$ð3ð ð3ð ð	3ð
 ð3ð Lð3ð ð3ð ð3ð ð3ð ˜D œKÑ'¨$Ñ.ð3ð ˜4 œ;Ñ&¨Ñ-ð3ð ˜jÑ(¨4Ñ/ð3ð 
ð3ð 3ð 3ð 3ð 3ð 3ð 3ð 3r   rV   )r   )r   )r   rJ   Útorchvision.transforms.v2r   ÚtvFÚimage_processing_backendsr   Úimage_processing_utilsr   Úimage_utilsr   r   r	   Úprocessing_utilsr
   r   Úutilsr   r   r   r   r-   r„   r3   rA   rT   rV   Ú__all__r   r   r   ú<module>r‘      sñ  ðð )Ð (à €€€Ø 7Ð 7Ð 7Ð 7Ð 7Ð 7à ;Ð ;Ð ;Ð ;Ð ;Ð ;Ø 2Ð 2Ð 2Ð 2Ð 2Ð 2Ø CÐ CÐ CÐ CÐ CÐ CÐ CÐ CÐ CÐ CØ 4Ð 4Ð 4Ð 4Ð 4Ð 4Ð 4Ð 4ðð ð ð ð ð ð ð ð
ð 
ð 
ð 
ð 
 ,°eð 
ñ 
ô 
ð 
ð ^bð-'ð -'Øð-'Ø$'ð-'Ø58ð-'ØKNð-'ØUZð-'à
ˆ3�ˆ8„_ð-'ð -'ð -'ð -'ð` Nð Àð Èð ð ð ð ð BCðð ØðØ+.ðØ;>ðà
Ð)Ô*ðð ð ð ð  ðKð Kð Kð Kð KÐ.ñ Kô Kñ „ðKð\ #Ð
#€€€r   