§
    l¬¸jîZ  ã                  óò   — d dl mZ d dlZd dlZd dlZd dlZd dlZd dlmZmZ d dl	m
Z
 d dlmZ d dlmZ d dlZd dlZd dlZd dlmZ d dlmZmZmZmZmZ d d	lmZmZmZmZm Z  d d
l!m"Z"  G d„ de¦  «        Z#dS )é    )ÚannotationsN)ÚcopyÚdeepcopy)Ú
ThreadPool)ÚPath)ÚAny)ÚDataset)ÚFORMATS_HELP_MSGÚHELP_URLÚIMG_FORMATSÚcheck_file_speedsÚget_split_fraction)ÚDEFAULT_CFGÚ
LOCAL_RANKÚLOGGERÚNUM_THREADSÚTQDM)Úimreadc                  óÔ   ‡ — e Zd ZdZ G d„ d¦  «        Zdddedddd	d
ddddfdAˆ fd$„ZdBd'„ZdCd*„Z	 dDdEd/„Z	dFd0„Z
dGd1„ZdHdId4„ZdJdId5„ZdFd6„ZdKd8„ZdKd9„ZdLd:„ZdMd<„ZdNdOd>„ZdPd@„Zˆ xZS )QÚBaseDataseta�
  Base dataset class for loading and processing image data.

    This class provides core functionality for loading images, caching, and preparing data for training and inference in
    object detection tasks.

    Attributes:
        img_path (str | list[str]): Path to the folder containing images.
        imgsz (int): Target image size for resizing.
        augment (bool): Whether to apply data augmentation.
        single_cls (bool): Whether to treat all objects as a single class.
        prefix (str): Prefix to print in log messages.
        fraction (float | int): Dataset ratio or image count to use.
        channels (int): Number of channels in the images (1 for grayscale, 3 for color). Color images loaded with OpenCV
            are in BGR channel order.
        cv2_flag (int): OpenCV flag for reading images.
        im_files (list[str]): List of image file paths.
        labels (list[dict]): List of label data dictionaries.
        ni (int): Number of images in the dataset.
        rect (bool): Whether to use rectangular training.
        batch_size (int): Size of batches.
        stride (int): Stride used in the model.
        pad (float): Padding value.
        buffer (list): Buffer for mosaic images.
        max_buffer_length (int): Maximum buffer size.
        ims (list): List of loaded images.
        im_hw0 (list): List of original image dimensions (h, w).
        im_hw (list): List of resized image dimensions (h, w).
        npy_files (list[Path]): List of numpy file paths.
        cache (str | None): Cache setting ('ram', 'disk', or None for no caching).
        transforms (callable): Image transformation function.
        batch_shapes (np.ndarray): Batch shapes for rectangular training.
        batch (np.ndarray): Batch index of each image.

    Methods:
        get_img_files: Read image files from the specified path.
        update_labels: Update labels to include only specified classes.
        load_image: Load an image from the dataset.
        cache_images: Cache images to memory or disk.
        cache_images_to_disk: Save an image as an *.npy file for faster loading.
        check_cache_disk: Check image caching requirements vs available disk space.
        check_cache_ram: Check image caching requirements vs available memory.
        set_rectangle: Sort images by aspect ratio and set batch shapes for rectangular training.
        get_image_and_label: Get and return label information from the dataset.
        update_labels_info: Custom label format method to be implemented by subclasses.
        build_transforms: Build transformation pipeline to be implemented by subclasses.
        get_labels: Get labels method to be implemented by subclasses.
    c                  ó2   — e Zd ZdZdd„Zdd	„Zdd„Zdd„ZdS )úBaseDataset._ImageCachezfStore images in one tensor, shared copy-on-write by fork workers and as shared memory by pickled ones.Úimagesúlist[np.ndarray]c                ó‚  — t          j        d„ |D ¦   «         ¦  «        | _        t          j        d„ |D ¦   «         ¦  «        | _        t          j        dgt          j        d„ |D ¦   «         ¦  «        f¦  «        | _        t          j        t          | j        d         ¦  «        t          j
        ¬¦  «        | _        | j                             ¦   «         }t          |¦  «        D ]Z\  }}|                     d¦  «                             t           j
        ¦  «        || j        |         | j        |dz            …<   d||<   Œ[dS )	zQPack images into one contiguous uint8 tensor and their layouts into NumPy arrays.c                ó   — g | ]	}|j         ‘Œ
S © ©Úshape©Ú.0Úims     úe/root/private/project/pallet_detection_v2/.venv/lib/python3.11/site-packages/ultralytics/data/base.pyú
<listcomp>z4BaseDataset._ImageCache.__init__.<locals>.<listcomp>O   s   € Ð#>Ð#>Ð#>° B¤HÐ#>Ð#>Ð#>ó    c                ó&   — g | ]}|j         j        ‘ŒS r   )ÚdtypeÚstrr    s     r#   r$   z4BaseDataset._ImageCache.__init__.<locals>.<listcomp>P   s   € Ð#BÐ#BÐ#B°R B¤H¤LÐ#BÐ#BÐ#Br%   r   c                ó   — g | ]	}|j         ‘Œ
S r   )Únbytesr    s     r#   r$   z4BaseDataset._ImageCache.__init__.<locals>.<listcomp>Q   s   € Ð:VÐ:VÐ:VÈ¸2¼9Ð:VÐ:VÐ:Vr%   éÿÿÿÿ)r'   é   N)ÚnpÚarrayÚshapesÚdtypesÚconcatenateÚcumsumÚoffsetsÚtorchÚemptyÚintÚuint8ÚbufferÚnumpyÚ	enumerateÚreshapeÚview)Úselfr   r8   Úir"   s        r#   Ú__init__z BaseDataset._ImageCache.__init__M   s  € åœ(Ð#>Ð#>°vÐ#>Ñ#>Ô#>Ñ?Ô?ˆDŒKÝœ(Ð#BÐ#B¸6Ð#BÑ#BÔ#BÑCÔCˆDŒKÝœ>¨A¨3µ´	Ð:VÐ:VÈvÐ:VÑ:VÔ:VÑ0WÔ0WÐ*XÑYÔYˆDŒLÝœ+¥c¨$¬,°rÔ*:Ñ&;Ô&;Å5Ä;ÐOÑOÔOˆDŒKØ”[×&Ò&Ñ(Ô(ˆFÝ" 6Ñ*Ô*ð !ð !‘��2Ø@BÇ
Â
È2ÁÄ×@SÒ@SÕTVÔT\Ñ@]Ô@]��t”| A”¨¬°a¸!±eÔ)<Ð<Ñ=Ø ��q‘	�	ð!ð !r%   r>   r6   Úreturnú
np.ndarrayc                ó<  — t          t          | j        ¦  «        ¦  «        |         }| j                             ¦   «         | j        |         | j        |dz            …                              | j        |         ¦  «                             | j        |         ¦  «        S )zReturn an image view by index.r,   )	ÚrangeÚlenr/   r8   r9   r3   r<   r0   r;   )r=   r>   s     r#   Ú__getitem__z#BaseDataset._ImageCache.__getitem__X   s   € å•c˜$œ+Ñ&Ô&Ñ'Ô'¨Ô*ˆAà”×!Ò!Ñ#Ô# D¤L°¤O°d´lÀ1ÀqÁ5Ô6IÐ$IÔJ×OÒOÐPTÔP[Ð\]ÔP^Ñ_Ô_×gÒgÐhlÔhsÐtuÔhvÑwÔwðr%   údict[str, Any]c                ó�  — | j                              ¦   «         }t          d¦  «        }| j                             ¦   «         s„|                     ¦   «         rpt          j        |¦  «        j        d| j	        d         z  k     rEt          j        |› d| j	        d         dz  d›d�¦  «         | j                             ¦   «         |d<   |S )	ziPickle the buffer by value, as each worker's own copy, when Linux shared memory is too small to share it.z/dev/shmé   r+   z too small to share é   @ú.1fzGB image cache, copying itr8   )Ú__dict__r   r   r8   Ú	is_sharedÚis_dirÚshutilÚ
disk_usageÚfreer3   r   Úwarningr9   )r=   ÚstateÚshms      r#   Ú__getstate__z$BaseDataset._ImageCache.__getstate___   s¼   € à”M×&Ò&Ñ(Ô(ˆEÝ�zÑ"Ô"ˆCØ”;×(Ò(Ñ*Ô*ð 6¨s¯zªz©|¬|ð 6ÅÔ@QÐRUÑ@VÔ@VÔ@[Ð^_ÐbfÔbnÐoqÔbrÑ^rÒ@rÐ@rÝ” #ÐwÐw¸4¼<ÈÔ;KÈwÑ;WÐwÐwÐwÐwÑxÔxÐxØ"&¤+×"3Ò"3Ñ"5Ô"5��h‘ØˆLr%   rR   c                ót   — t          j        |d         ¦  «        |d<   | j                             |¦  «         dS )zRestore the buffer as a tensor.r8   N)r4   Ú	as_tensorrK   Úupdate)r=   rR   s     r#   Ú__setstate__z$BaseDataset._ImageCache.__setstate__h   s5   € å#œo¨e°H¬oÑ>Ô>ˆE�(‰OØŒM× Ò  Ñ'Ô'Ð'Ð'Ð'r%   N)r   r   )r>   r6   r@   rA   )r@   rF   )rR   rF   )Ú__name__Ú
__module__Ú__qualname__Ú__doc__r?   rE   rT   rX   r   r%   r#   Ú_ImageCacher   J   sj   € € € € € ØtÐtð		!ð 		!ð 		!ð 		!ð	ð 	ð 	ð 	ð	ð 	ð 	ð 	ð	(ð 	(ð 	(ð 	(ð 	(ð 	(r%   r]   i€  FTÚ é   é    g      à?Nç      ð?é   Úimg_pathústr | list[str]Úimgszr6   Úcacheú
bool | strÚaugmentÚboolÚhyprF   Úprefixr(   ÚrectÚ
batch_sizeÚstrideÚpadÚfloatÚ
single_clsÚclassesúlist[int] | NoneÚfractionÚchannelsc                ó  •— t          ¦   «                              ¦   «          || _        || _        || _        || _        || _        t          |d¦  «        | _        || _	        |dk    rt          j        nt          j        | _        |                      | j        ¦  «        | _        |                      ¦   «         | _        |                      |¬¦  «         t'          | j        ¦  «        | _        || _        || _        |	| _        |
| _        | j        r| j        €J ‚|                      ¦   «          g | _        | j        rt7          | j        | j        dz  df¦  «        nd| _        dg| j        z  dg| j        z  dg| j        z  c| _        | _        | _        d„ | j        D ¦   «         | _         tC          |tD          ¦  «        r| #                    ¦   «         n|d	u rd
nd| _$        | j$        d
k    rD|  %                    ¦   «         r0|j&        rtO          j(        d¦  «         |  )                    ¦   «          n3| j$        dk    r(|  *                    ¦   «         r|  )                    ¦   «          |  +                    tY          |¦  «        ¬¦  «        | _-        dS )aK  Initialize BaseDataset with given configuration and options.

        Args:
            img_path (str | list[str]): Path to the folder containing images or list of image paths.
            imgsz (int): Image size for resizing.
            cache (bool | str): Cache images to RAM or disk during training.
            augment (bool): If True, data augmentation is applied.
            hyp (dict[str, Any]): Hyperparameters to apply data augmentation.
            prefix (str): Prefix to print in log messages.
            rect (bool): If True, rectangular training is used.
            batch_size (int): Size of batches.
            stride (int): Stride used in the model.
            pad (float): Padding value.
            single_cls (bool): If True, single class training is used.
            classes (list[int], optional): List of included classes.
            fraction (float | int): Dataset ratio or image count to use.
            channels (int): Number of channels in the images (1 for grayscale, 3 for color). Color images loaded with
                OpenCV are in BGR channel order.
        Útrainr,   )Úinclude_classNé   iè  r   c                óR   — g | ]$}t          |¦  «                             d ¦  «        ‘Œ%S )z.npy)r   Úwith_suffix)r!   Úfs     r#   r$   z(BaseDataset.__init__.<locals>.<listcomp>¬   s,   € ÐMÐMÐM¸!�$˜q™'œ'×-Ò-¨fÑ5Ô5ÐMÐMÐMr%   TÚramz‹cache='ram' may produce non-deterministic training results. Consider cache='disk' as a deterministic alternative if your disk space allows.Údisk)rj   ).Úsuperr?   rc   re   rh   rq   rk   r   rt   ru   Úcv2ÚIMREAD_GRAYSCALEÚIMREAD_COLORÚcv2_flagÚget_img_filesÚim_filesÚ
get_labelsÚlabelsÚupdate_labelsrD   Únirl   rm   rn   ro   Úset_rectangler8   ÚminÚmax_buffer_lengthÚimsÚim_hw0Úim_hwÚ	npy_filesÚ
isinstancer(   Úlowerrf   Úcheck_cache_ramÚdeterministicr   rQ   Úcache_imagesÚcheck_cache_diskÚbuild_transformsr   Ú
transforms)r=   rc   re   rf   rh   rj   rk   rl   rm   rn   ro   rq   rr   rt   ru   Ú	__class__s                  €r#   r?   zBaseDataset.__init__m   si  ø€ õH 	‰Œ×ÒÑÔÐØ ˆŒØˆŒ
ØˆŒØ$ˆŒØˆŒÝ*¨8°WÑ=Ô=ˆŒØ ˆŒØ08¸A²°�Ô,Ð,Å3ÔCSˆŒØ×*Ò*¨4¬=Ñ9Ô9ˆŒØ—o’oÑ'Ô'ˆŒØ×Ò¨ÐÑ1Ô1Ð1Ý�d”kÑ"Ô"ˆŒØˆŒ	Ø$ˆŒØˆŒØˆŒØŒ9ð 	!Ø”?Ð.Ð.Ð.Ø×ÒÑ Ô Ð ð ˆŒØNRÌlÐ!a¥ d¤g¨t¬ÀÑ/BÀDÐ%IÑ!JÔ!JÐ!JÐ`aˆÔð .2¨F°T´WÑ,<¸t¸fÀtÄwÑ>NÐQUÐPVÐY]ÔY`ÑP`Ð)ˆŒ�$”+˜tœzØMÐM¸t¼}ÐMÑMÔMˆŒÝ&0°½Ñ&<Ô&<Ðb�U—[’[‘]”]�]È5ÐTXÈ=È=À%À%Ð^bˆŒ
ØŒ:˜ÒÐ 4×#7Ò#7Ñ#9Ô#9ÐØÔ ð Ý”ðfñô ð ð ×ÒÑÔÐÐØŒZ˜6Ò!Ð! d×&;Ò&;Ñ&=Ô&=Ð!Ø×ÒÑÔÐð ×/Ò/µD¸±I´IÐ/Ñ>Ô>ˆŒˆˆr%   r@   ú	list[str]c                ó’  ‡— 	 g }t          |t          ¦  «        r|n|gD �]2}t          |¦  «        }|                     ¦   «         rL|t	          j        t          t          t	          j        |¦  «        ¦  «        dz  dz  ¦  «        d¬¦  «        z  }Œr|                     ¦   «         r•t          |d¬¦  «        5 }| 	                    ¦   «          
                    ¦   «                              ¦   «         }t          |j        ¦  «        t          j        z   Š|ˆfd„|D ¦   «         z  }ddd¦  «         n# 1 swxY w Y   �Œt          | j        › |› d	�¦  «        ‚t#          d
„ |D ¦   «         ¦  «        }|sJ | j        › d|› dt$          › �¦   «         ‚n4# t&          $ r'}t          | j        › d|› dt(          › �¦  «        |‚d}~ww xY wt          | j        t,          ¦  «        r| j        n1t/          dt1          t3          |¦  «        | j        z  ¦  «        ¦  «        }|t3          |¦  «        k     r
|d|…         n|}t5          || j        ¬¦  «         |S )aE  Read image files from the specified path.

        Args:
            img_path (str | list[str]): Path or list of paths to image directories or files.

        Returns:
            (list[str]): List of image file paths.

        Raises:
            FileNotFoundError: If no images are found or the path doesn't exist.
        z**z*.*T)Ú	recursivezutf-8)Úencodingc                ól   •— g | ]0}|                      d ¦  «        r|                     d ‰d¦  «        n|‘Œ1S )z./r,   )Ú
startswithÚreplace)r!   ÚxÚparents     €r#   r$   z-BaseDataset.get_img_files.<locals>.<listcomp>Ò   sA   ø€ ÐaÐaÐaÐZ[¸A¿LºLÈÑ<NÔ<NÐU˜aŸiši¨¨f°aÑ8Ô8Ð8ÐTUÐaÐaÐar%   Nz does not existc              3  ó¾   K  — | ]X}|                      d ¦  «        d                              ¦   «         t          v ¯6|                     dt          j        ¦  «        V — ŒYdS )ú.r+   ú/N)Ú
rpartitionr’   r   r    ÚosÚsep©r!   r¡   s     r#   ú	<genexpr>z,BaseDataset.get_img_files.<locals>.<genexpr>Ö   s\   è è € ÐpÐp¸À1Ç<Â<ÐPSÑCTÔCTÐUWÔCX×C^ÒC^ÑC`ÔC`ÕdoÐCoÐCo˜aŸiši¨­R¬VÑ4Ô4ÐCoÐCoÐCoÐCoÐpÐpr%   zNo images found in z. zError loading data from ú
r,   )rk   )r‘   Úlistr   rM   Úglobr(   ÚescapeÚis_fileÚopenÚreadÚstripÚ
splitlinesr¢   r§   r¨   ÚFileNotFoundErrorrk   Úsortedr
   Ú	Exceptionr   rt   r6   ÚmaxÚroundrD   r   )	r=   rc   r|   ÚpÚtr…   ÚeÚcountr¢   s	           @r#   r„   zBaseDataset.get_img_files»   s¢  ø€ ð	kØˆAÝ!+¨HµdÑ!;Ô!;ÐK�X�XÀ(Àð Pñ P�Ý˜‘G”G�Ø—8’8‘:”:ð 
PØ�œ¥3¥t­D¬K¸©N¬NÑ';Ô';¸dÑ'BÀUÑ'JÑ#KÔ#KÐW[Ð\Ñ\Ô\Ñ\�A�Aà—Y’Y‘[”[ð PÝ˜a¨'Ð2Ñ2Ô2ð b°aØŸFšF™HœHŸNšNÑ,Ô,×7Ò7Ñ9Ô9˜Ý!$ Q¤X¡¤µ´Ñ!7˜ØÐaÐaÐaÐaÐ_`ÐaÑaÔaÑa˜ðbð bð bñ bô bð bð bð bð bð bð bøøøð bð bð bð bùõ ,¨t¬{Ð,N¸AÐ,NÐ,NÐ,NÑOÔOÐOÝÐpÐp¸aÐpÑpÔpÑpÔpˆHàÐ^Ð^ ¤Ð^Ð^ÀÐ^Ð^ÕL\Ð^Ð^Ñ^Ô^�8Ð^øÝð 	kð 	kð 	kÝ# t¤{Ð$bÐ$bÈHÐ$bÐ$bÕX`Ð$bÐ$bÑcÔcÐijÐjøøøøð	køøøå!+¨D¬M½3Ñ!?Ô!?Ðq�”�ÅSÈÍEÕRUÐV^ÑR_ÔR_ÐbfÔboÑRoÑLpÔLpÑEqÔEqˆØ',­s°8©}¬}Ò'<Ð'<�8˜F˜U˜FÔ#Ð#À(ˆÝ˜(¨4¬;Ð7Ñ7Ô7Ð7Øˆs>   ƒB3F Â6A+D-Ä!F Ä-D1	Ä1F Ä4D1	Ä5AF Æ
F<Æ"F7Æ7F<rx   ÚNonec                ó¨  ‡— t          j        |¦  «                             dd¦  «        }t          t	          | j        ¦  «        ¦  «        D �]}|�á| j        |         d         }| j        |         d         }| j        |         d         Š| j        |                              d¦  «        }||k                         d¦  «        }||         | j        |         d<   ||         | j        |         d<   ‰r)ˆfd„t          |¦  «        D ¦   «         | j        |         d<   |�||         | j        |         d<   | j	        rd	| j        |         d         dd…<   �ŒdS )
z¸Update labels to include only specified classes.

        Args:
            include_class (list[int], optional): List of classes to include. If None, all classes are included.
        r,   r+   NÚclsÚbboxesÚsegmentsÚ	keypointsc                ó*   •— g | ]\  }}|¯‰|         ‘ŒS r   r   )r!   ÚsiÚidxrÁ   s      €r#   r$   z-BaseDataset.update_labels.<locals>.<listcomp>ñ   s'   ø€ Ð1bÐ1bÐ1bÁ7À2ÀsÐ^aÐ1b°(¸2´,Ð1bÐ1bÐ1br%   r   )
r-   r.   r;   rC   rD   r‡   ÚgetÚanyr:   rq   )	r=   rx   Úinclude_class_arrayr>   r¿   rÀ   rÂ   ÚjrÁ   s	           @r#   rˆ   zBaseDataset.update_labelsà   sR  ø€ õ !œh }Ñ5Ô5×=Ò=¸aÀÑDÔDÐÝ•s˜4œ;Ñ'Ô'Ñ(Ô(ð 	-ñ 	-ˆAØÐ(Ø”k !”n UÔ+�Øœ Qœ¨Ô1�Øœ; qœ>¨*Ô5�Ø œK¨œN×.Ò.¨{Ñ;Ô;�	ØÐ/Ò/×4Ò4°QÑ7Ô7�Ø(+¨A¬�”˜A”˜uÑ%Ø+1°!¬9�”˜A”˜xÑ(Øð cØ1bÐ1bÐ1bÐ1bÍiÐXYÉlÌlÐ1bÑ1bÔ1b�D”K ”N :Ñ.ØÐ(Ø2;¸A´,�D”K ”N ;Ñ/ØŒð -Ø+,�”˜A”˜uÔ% a a aÑ(ùð	-ð 	-r%   r>   Ú	rect_modeÚresize_shortú3tuple[np.ndarray, tuple[int, int], tuple[int, int]]c           	     óü  — | j         |         | j        |         | j        |         }}}|�€­|                     ¦   «         �r	 t	          j        |¦  «        }|j        dk    r|j        d         nd}|| j        k    rbt          j
        | j        › d|› d|› d| j        › �¦  «         t          |¦  «                             d¬	¦  «         t          || j        ¬
¦  «        }n‡# t           $ rd}t          j
        | j        › d|› d|› �¦  «         t          |¦  «                             d¬	¦  «         t          || j        ¬
¦  «        }Y d}~nd}~ww xY wt          || j        ¬
¦  «        }|€t#          d|› �¦  «        ‚|j        dd…         \  }	}
|�r|r‡| j        t'          |	|
¦  «        z  }|dk    rh|	|
k     rt)          j        |
|z  ¦  «        | j        fn| j        t)          j        |	|z  ¦  «        f\  }}t-          j        |||ft,          j        ¬¦  «        }n×| j        t3          |	|
¦  «        z  }|dk    rwt'          t)          j        |
|z  ¦  «        | j        ¦  «        t'          t)          j        |	|z  ¦  «        | j        ¦  «        }}t-          j        |||ft,          j        ¬¦  «        }nA|	|
cxk    r| j        k    s/n t-          j        || j        | j        ft,          j        ¬¦  «        }|j        dk    r|d         }| j        r½| j        dk    r²||	|
f|j        dd…         c| j         |<   | j        |<   | j        |<   | j                             |¦  «         dtA          | j        ¦  «        cxk     r| j!        k    rFn nC| j         "                    d¦  «        }| j        dk    rd\  | j         |<   | j        |<   | j        |<   ||	|
f|j        dd…         fS | j         |         | j        |         | j        |         fS )a¾  Load an image from dataset index 'i'.

        Args:
            i (int): Index of the image to load.
            rect_mode (bool): Whether to use rectangular resizing (long side to imgsz).
            resize_short (bool): Whether to resize the shorter side to imgsz while maintaining aspect ratio. Overrides
                rect_mode when True.

        Returns:
            im (np.ndarray): Loaded image as a NumPy array.
            hw_original (tuple[int, int]): Original image dimensions in (height, width) format.
            hw_resized (tuple[int, int]): Resized image dimensions in (height, width) format.

        Raises:
            FileNotFoundError: If the image file is not found.
        Nrb   r+   r,   z Removing stale *.npy image file z with z channels, expected T©Ú
missing_ok©Úflagsz"Removing corrupt *.npy image file z	 due to: zImage Not Found rH   )Úinterpolation).Nr}   r   )NNN)#r�   r…   r�   Úexistsr-   ÚloadÚndimr   ru   r   rQ   rk   r   Úunlinkr   rƒ   r¶   r´   re   r‹   ÚmathÚceilr€   ÚresizeÚINTER_LINEARr·   rh   rf   rŽ   r�   r8   ÚappendrD   rŒ   Úpop)r=   r>   rÊ   rË   r"   r|   ÚfnÚnpy_channelsr»   Úh0Úw0ÚrÚwÚhrÉ   s                  r#   Ú
load_imagezBaseDataset.load_image÷   s2  € ð& ”H˜Q”K ¤¨qÔ!1°4´>À!Ô3DˆrˆAˆØ‰:Ø�yŠy‰{Œ{ñ 4ð8Ýœ ™œ�BØ35´7¸a²<°< 2¤8¨B¤< <ÀQ�LØ# t¤}Ò4Ð4ÝœØ#œ{ð  Hð  HÈBð  Hð  HÐVbð  Hð  HÐx|ô  yFð  Hð  Hñô ð õ ˜R™œŸš°4˜Ñ8Ô8Ð8Ý# A¨T¬]Ð;Ñ;Ô;˜øøÝ ð 8ð 8ð 8Ý”N d¤kÐ#eÐ#eÐUWÐ#eÐ#eÐbcÐ#eÐ#eÑfÔfÐfÝ˜‘H”H—O’O¨t�OÑ4Ô4Ð4Ý ¨¬Ð7Ñ7Ô7�B�B�B�B�B�Bøøøøð8øøøõ
 ˜A T¤]Ð3Ñ3Ô3�ØˆzÝ'Ð(>¸1Ð(>Ð(>Ñ?Ô?Ð?à”X˜b˜q˜b”\‰FˆB�Øñ ^Øð 	TØœ
¥S¨¨R¡[¤[Ñ0�AØ˜A’v�vØBDÀrÂ'À'¥¤	¨"¨q©&Ñ 1Ô 1°4´:Ð>Ð>ÐPTÔPZÕ\`Ô\eÐfhÐklÑflÑ\mÔ\mÐOn™˜˜1Ý œZ¨¨Q°¨FÅ#ÔBRÐSÑSÔS˜øàœ
¥S¨¨R¡[¤[Ñ0�AØ˜A’v�vÝ #¥D¤I¨b°1©fÑ$5Ô$5°t´zÑ BÔ BÅCÍÌ	ÐRTÐWXÑRXÑHYÔHYÐ[_Ô[eÑDfÔDf˜1˜Ý œZ¨¨Q°¨FÅ#ÔBRÐSÑSÔS˜øØ˜BÐ,Ð,Ò,Ð, $¤*Ò,Ð,Ð,Ð,Ý”Z  T¤Z°´Ð$<ÍCÔL\Ð]Ñ]Ô]�ØŒw˜!Š|ˆ|Ø˜	”]�ð Œ|ð V ¤
¨eÒ 3Ð 3Ø=?À"ÀbÀÈ2Ì8ÐTVÐUVÐTVÌ<Ð:�”˜‘˜Tœ[¨™^¨T¬Z¸©]Ø”×"Ò" 1Ñ%Ô%Ð%Ø•s˜4œ;Ñ'Ô'ÐAÐAÒAÐA¨4Ô+AÒAÐAÐAÐAÐAØœŸš¨Ñ*Ô*�AØ”z UÒ*Ð*ØEUÑB˜œ ™ T¤[°¡^°T´ZÀ±]à˜˜B�x ¤¨"¨1¨"¤Ð-Ð-àŒx˜Œ{˜DœK¨œN¨D¬J°q¬MÐ9Ð9s   ÁBC Ã
EÃ'AEÅEc                ó"  — d\  }}| j         dk    r	| j        dfn| j        df\  }}t          t          ¦  «        5 }|                     |t          | j        ¦  «        ¦  «        }t          t          |¦  «        | j        t          dk    ¬¦  «        }|D ]¥\  }}	| j         dk    rI|| j        |                              ¦   «         r$| j        |                              ¦   «         j        ndz  }n3|	\  | j        |<   | j        |<   | j        |<   || j        |         j        z  }| j        › d||z  d›d	|› d
�|_        Œ¦|                     ¦   «          ddd¦  «         n# 1 swxY w Y   | j         dk    r!|                      | j        ¦  «        | _        dS dS )z3Cache images to memory or disk for faster training.©r   rI   r~   ÚDiskÚRAMr   )ÚtotalÚdisablezCaching images (rJ   zGB ú)Nr}   )rf   Úcache_images_to_diskrä   r   r   ÚimaprC   r‰   r   r:   r   r�   rÓ   ÚstatÚst_sizer�   rŽ   r�   r*   rk   ÚdescÚcloser]   )
r=   ÚbÚgbÚfcnÚstorageÚpoolÚresultsÚpbarr>   r¡   s
             r#   r•   zBaseDataset.cache_images=  sÒ  € à‰ˆˆ2Ø>B¼jÈFÒ>RÐ>R˜Ô1°6Ð:Ð:ÐY]ÔYhÐjoÐXp‰ˆˆWÝ�Ñ$Ô$ð 
	¨Ø—i’i ¥U¨4¬7¡^¤^Ñ4Ô4ˆGÝ�	 'Ñ*Ô*°$´'Å:ÐPQÂ>ÐRÑRÔRˆDØð Vð V‘��1Ø”: Ò'Ð'Ø¸T¼^ÈAÔ=N×=UÒ=UÑ=WÔ=WÐ^˜œ¨Ô*×/Ò/Ñ1Ô1Ô9Ð9Ð]^Ñ^�A�AàABÑ>�D”H˜Q‘K ¤¨Q¡°´¸A±Ø˜œ !œÔ+Ñ+�AØ#œ{ÐUÐU¸AÀ¹FÐUÐUÐUÈ7ÐUÐUÐU�”	�	Ø�JŠJ‰LŒLˆLð
	ð 
	ð 
	ñ 
	ô 
	ð 
	ð 
	ð 
	ð 
	ð 
	ð 
	øøøð 
	ð 
	ð 
	ð 
	ð Œ:˜ÒÐØ×'Ò'¨¬Ñ1Ô1ˆDŒHˆHˆHð Ðs   ºDEÅEÅEc                ó|  — | j         |         }|                     ¦   «         sš	 t          j        |                     ¦   «         t          | j        |         | j        ¬¦  «        d¬¦  «         dS # t          $ rB}| 	                    d¬¦  «         t          j        | j        › d|› d|› �¦  «         Y d}~dS d}~ww xY wdS )	z2Save an image as an *.npy file for faster loading.rÐ   F)Úallow_pickleTrÎ   u%   WARNING âš ï¸� Failed to cache image z: N)r�   rÓ   r-   ÚsaveÚas_posixr   r…   rƒ   r¶   rÖ   r   rQ   rk   )r=   r>   r|   r»   s       r#   rì   z BaseDataset.cache_images_to_diskO  sà   € àŒN˜1ÔˆØ�xŠx‰zŒzð 	^ð^Ý”˜Ÿ
š
™œ¥f¨T¬]¸1Ô-=ÀTÄ]Ð&SÑ&SÔ&SÐbgÐhÑhÔhÐhÐhÐhøÝð ^ð ^ð ^Ø—’ D�Ñ)Ô)Ð)Ý” $¤+Ð\Ð\ÐTUÐ\Ð\ÐYZÐ\Ð\Ñ]Ô]Ð]Ð]Ð]Ð]Ð]Ð]Ð]øøøøð^øøøð	^ð 	^s   £AA- Á-
B9Á77B4Â4B9çš™™™™™¹?Úsafety_marginc                óª  — d\  }}t          | j        d¦  «        }t          |¦  «        D ]Ž}t          j        | j        ¦  «        }t          |¦  «        }|€Œ-||j        z  }t          j	        t          |¦  «        j        t          j        ¦  «        s&d| _        t          j        | j        › d�¦  «          dS Œ�|| j        z  |z  d|z   z  }t#          j        t          | j        d         ¦  «        j        ¦  «        \  }	}
}||k    rLd| _        t          j        | j        › ||z  d›d	t'          |d
z  ¦  «        › d||z  d›d|	|z  d›d�	¦  «         dS dS )züCheck if there's enough disk space for caching images.

        Args:
            safety_margin (float): Safety margin factor for disk space calculation.

        Returns:
            (bool): True if there's enough disk space, False otherwise.
        ræ   é   Nz7Skipping caching images to disk, directory not writableFr,   r   rJ   zGB disk space required, with éd   ú% safety margin but only r¥   z#GB free, not caching images to diskT)r‹   r‰   rC   ÚrandomÚchoicer…   r   r*   r§   Úaccessr   r¢   ÚW_OKrf   r   rQ   rk   rN   rO   r6   )r=   rþ   rò   ró   ÚnÚ_Úim_filer"   Údisk_requiredré   Ú_usedrP   s               r#   r–   zBaseDataset.check_cache_diskY  s”  € ð ‰ˆˆ2Ý�”˜ÑÔˆÝ�q‘”ð 		ð 		ˆAÝ”m D¤MÑ2Ô2ˆGÝ˜‘”ˆBØˆzØØ�”‰NˆAÝ”9�T '™]œ]Ô1µ2´7Ñ;Ô;ð Ø!�”
Ý” $¤+ÐfÐfÐfÑgÔgÐgØ�u�uðð ˜DœG™ a™¨1¨}Ñ+<Ñ=ˆÝ#Ô.­t°D´MÀ!Ô4DÑ/EÔ/EÔ/LÑMÔMÑˆˆu�dØ˜4ÒÐØˆDŒJÝŒNØ”;ð W °Ñ 2ÐPð Wð WÝ˜M¨CÑ/Ñ0Ô0ðWð Wà˜"‘9ÐVðWð Wà#(¨2¡:ÐVðWð Wð Wñô ð ð
 �5Øˆtr%   c                óü  — d\  }}t          | j        d¦  «        }t          |¦  «        D ]<}||                      t	          j        | j        ¦  «        ¦  «        d         j        z  }Œ=|| j        z  |z  d|z   z  }t          d¦  «                             ¦   «         }||j	        k    rVd| _
        t          j        | j        › ||z  d›dt          |d	z  ¦  «        › d
|j	        |z  d›d|j        |z  d›d�	¦  «         dS dS )zçCheck if there's enough RAM for caching images.

        Args:
            safety_margin (float): Safety margin factor for RAM calculation.

        Returns:
            (bool): True if there's enough RAM, False otherwise.
        ræ   r   r   r,   ÚpsutilNrJ   z%GB RAM required to cache images with r  r  r¥   z GB available, not caching imagesFT)r‹   r‰   rC   rä   r  Ú	randranger*   Ú
__import__Úvirtual_memoryÚ	availablerf   r   rQ   rk   r6   ré   )r=   rþ   rò   ró   r  r  Úmem_requiredÚmems           r#   r“   zBaseDataset.check_cache_ramz  s<  € ð ‰ˆˆ2Ý�”˜ÑÔˆÝ�q‘”ð 	Fð 	FˆAØ�—’¥Ô!1°$´'Ñ!:Ô!:Ñ;Ô;¸AÔ>ÔEÑEˆAˆAØ˜4œ7‘{ Q‘¨!¨mÑ*;Ñ<ˆÝ˜Ñ"Ô"×1Ò1Ñ3Ô3ˆØ˜#œ-Ò'Ð'ØˆDŒJÝŒNØ”;ð a ¨rÑ 1ÐWð að aÝ˜M¨CÑ/Ñ0Ô0ðað aà”= 2Ñ%Ð`ðað aà,/¬I¸©NÐ`ðað að añô ð ð
 �5Øˆtr%   c                ó\  ‡ — t          j        t          j        ‰ j        ¦  «        ‰ j        z  ¦  «                             t          ¦  «        }|d         dz   }t          j        d„ ‰ j        D ¦   «         ¦  «        }|dd…df         |dd…df         z  }| 	                    ¦   «         }ˆ fd„|D ¦   «         ‰ _
        ˆ fd„|D ¦   «         ‰ _        ||         }ddgg|z  }t          |¦  «        D ]T}|||k             }|                     ¦   «         |                     ¦   «         }
}	|
dk     r|
dg||<   ŒD|	dk    r
dd|	z  g||<   ŒUt          j        t          j        |¦  «        ‰ j        z  ‰ j        z  ‰ j        z   ¦  «                             t          ¦  «        ‰ j        z  ‰ _        |‰ _        dS )zJSort images by aspect ratio and set batch shapes for rectangular training.r+   r,   c                ó8   — g | ]}|                      d ¦  «        ‘ŒS r   )rÜ   r©   s     r#   r$   z-BaseDataset.set_rectangle.<locals>.<listcomp>˜  s"   € Ð:Ð:Ð:¨�a—e’e˜G‘n”nÐ:Ð:Ð:r%   Nr   c                ó*   •— g | ]}‰j         |         ‘ŒS r   )r…   ©r!   r>   r=   s     €r#   r$   z-BaseDataset.set_rectangle.<locals>.<listcomp>›  s    ø€ Ð9Ð9Ð9¨a˜œ qÔ)Ð9Ð9Ð9r%   c                ó*   •— g | ]}‰j         |         ‘ŒS r   )r‡   r  s     €r#   r$   z-BaseDataset.set_rectangle.<locals>.<listcomp>œ  s   ø€ Ð5Ð5Ð5¨!�t”{ 1”~Ð5Ð5Ð5r%   )r-   ÚfloorÚaranger‰   rm   Úastyper6   r.   r‡   Úargsortr…   rC   r‹   r·   rØ   re   rn   ro   Úbatch_shapesÚbatch)r=   ÚbiÚnbÚsÚarÚirectr/   r>   ÚariÚminiÚmaxis   `          r#   rŠ   zBaseDataset.set_rectangle“  s£  ø€ åŒX•b”i ¤Ñ(Ô(¨4¬?Ñ:Ñ;Ô;×BÒBÅ3ÑGÔGˆØ�ŒV�a‰ZˆåŒHÐ:Ð:¨d¬kÐ:Ñ:Ô:Ñ;Ô;ˆØˆqˆqˆq�!ˆtŒW�q˜˜˜˜A˜”wÑˆØ—
’
‘”ˆØ9Ð9Ð9Ð9°5Ð9Ñ9Ô9ˆŒØ5Ð5Ð5Ð5¨uÐ5Ñ5Ô5ˆŒØ�ŒYˆð �a�&�˜B‘ˆÝ�r‘”ð 	*ð 	*ˆAØ�R˜1’W”+ˆCØŸš™œ C§G¢G¡I¤I�$ˆDØ�aŠxˆxØ! 1˜I��q‘	�	Ø˜’�Ø  D¡˜M��q‘	øåœG¥B¤H¨VÑ$4Ô$4°t´zÑ$AÀDÄKÑ$OÐRVÔRZÑ$ZÑ[Ô[×bÒbÕcfÑgÔgÐjnÔjuÑuˆÔØˆŒ
ˆ
ˆ
r%   Úindexc                óR   — |                       |                      |¦  «        ¦  «        S )z5Return transformed label information for given index.)r˜   Úget_image_and_label)r=   r'  s     r#   rE   zBaseDataset.__getitem__¬  s"   € à�Š˜t×7Ò7¸Ñ>Ô>Ñ?Ô?Ð?r%   c                óŽ  — t          | j        |         ¦  «        }|                     dd¦  «         |                      |¦  «        \  |d<   |d<   |d<   |d         d         |d         d         z  |d         d         |d         d         z  f|d<   | j        r| j        | j        |                  |d	<   |                      |¦  «        S )
zÝGet and return label information from the dataset.

        Args:
            index (int): Index of the image to retrieve.

        Returns:
            (dict[str, Any]): Label dictionary with image and metadata.
        r   NÚimgÚ	ori_shapeÚresized_shaper   r,   Ú	ratio_padÚ
rect_shape)r   r‡   rÜ   rä   rl   r  r  Úupdate_labels_info)r=   r'  Úlabels      r#   r)  zBaseDataset.get_image_and_label°  sÑ   € õ ˜œ UÔ+Ñ,Ô,ˆØ�	Š	�'˜4Ñ Ô Ð ØCGÇ?Â?ÐSXÑCYÔCYÑ@ˆˆe‰�e˜KÑ(¨%°Ñ*@à�/Ô" 1Ô%¨¨kÔ(:¸1Ô(=Ñ=Ø�/Ô" 1Ô%¨¨kÔ(:¸1Ô(=Ñ=ð
ˆˆkÑð Œ9ð 	GØ"&Ô"3°D´J¸uÔ4EÔ"FˆE�,ÑØ×&Ò& uÑ-Ô-Ð-r%   c                ó*   — t          | j        ¦  «        S )z5Return the length of the labels list for the dataset.)rD   r‡   ©r=   s    r#   Ú__len__zBaseDataset.__len__Ä  s   € å�4”;ÑÔÐr%   r1  c                ó   — |S )z!Customize your label format here.r   )r=   r1  s     r#   r0  zBaseDataset.update_labels_infoÈ  s   € àˆr%   údict[str, Any] | Nonec                ó   — t           ‚)a  Users can customize augmentations here.

        Examples:
            >>> if self.augment:
            ...     # Training transforms
            ...     return Compose([])
            >>> else:
            ...    # Val transforms
            ...    return Compose([])
        ©ÚNotImplementedError)r=   rj   s     r#   r—   zBaseDataset.build_transformsÌ  s
   € õ "Ð!r%   úlist[dict[str, Any]]c                ó   — t           ‚)a   Users can customize their own format here.

        Examples:
            Ensure output is a dictionary with the following keys:
            >>> dict(
            ...     im_file=im_file,
            ...     shape=shape,  # format: (height, width)
            ...     cls=cls,
            ...     bboxes=bboxes,  # xywh
            ...     segments=segments,  # xy
            ...     keypoints=keypoints,  # xy
            ...     normalized=True,  # or False
            ...     bbox_format="xyxy",  # or xywh, ltwh
            ... )
        r8  r3  s    r#   r†   zBaseDataset.get_labelsÙ  s
   € õ  "Ð!r%   )rc   rd   re   r6   rf   rg   rh   ri   rj   rF   rk   r(   rl   ri   rm   r6   rn   r6   ro   rp   rq   ri   rr   rs   rt   rp   ru   r6   )rc   rd   r@   rš   )rx   rs   r@   r½   )TF)r>   r6   rÊ   ri   rË   ri   r@   rÌ   )r@   r½   )r>   r6   r@   r½   )rý   )rþ   rp   r@   ri   )ra   )r'  r6   r@   rF   )r@   r6   )r1  rF   r@   rF   )N)rj   r6  )r@   r:  )rY   rZ   r[   r\   r]   r   r?   r„   rˆ   rä   r•   rì   r–   r“   rŠ   rE   r)  r4  r0  r—   r†   Ú__classcell__)r™   s   @r#   r   r      sê  ø€ € € € € ð.ð .ð`!(ð !(ð !(ð !(ð !(ñ !(ô !(ð !(ðL Ø!ØØ)ØØØØØØ Ø$(ØØðL?ð L?ð L?ð L?ð L?ð L?ð L?ð\#ð #ð #ð #ðJ-ð -ð -ð -ð0 DIðD:ð D:ð D:ð D:ð D:ðL2ð 2ð 2ð 2ð$^ð ^ð ^ð ^ðð ð ð ð ðBð ð ð ð ð2ð ð ð ð2@ð @ð @ð @ð.ð .ð .ð .ð( ð  ð  ð  ðð ð ð ð"ð "ð "ð "ð "ð"ð "ð "ð "ð "ð "ð "ð "r%   r   )$Ú
__future__r   r­   r×   r§   r  rN   r   r   Úmultiprocessing.poolr   Úpathlibr   Útypingr   r€   r9   r-   r4   Útorch.utils.datar	   Úultralytics.data.utilsr
   r   r   r   r   Úultralytics.utilsr   r   r   r   r   Úultralytics.utils.patchesr   r   r   r%   r#   ú<module>rE     sh  ðð #Ð "Ð "Ð "Ð "Ð "à €€€Ø €€€Ø 	€	€	€	Ø €€€Ø €€€Ø Ð Ð Ð Ð Ð Ð Ð Ø +Ð +Ð +Ð +Ð +Ð +Ø Ð Ð Ð Ð Ð Ø Ð Ð Ð Ð Ð à 
€
€
€
Ø Ð Ð Ð Ø €€€Ø $Ð $Ð $Ð $Ð $Ð $à qÐ qÐ qÐ qÐ qÐ qÐ qÐ qÐ qÐ qÐ qÐ qÐ qÐ qØ PÐ PÐ PÐ PÐ PÐ PÐ PÐ PÐ PÐ PÐ PÐ PÐ PÐ PØ ,Ð ,Ð ,Ð ,Ð ,Ð ,ðP"ð P"ð P"ð P"ð P"�'ñ P"ô P"ð P"ð P"ð P"r%   