§
    à©ºj¥'  ã                  óŒ   — d Z ddlmZ ddlZddlmZ ddlZddlm	Z
 ddlmZmZmZ  G d„ d	¦  «        Z G d
„ d¦  «        Zd„ ZdS )u³  PyTorch ecosystem integrations for hotcoco.

Provides ``CocoEvaluator`` and ``CocoDetection`` as drop-in replacements for
the equivalent classes in torchvision's detection reference scripts. No
torchvision or pycocotools dependency is required.

PyTorch and Pillow are optional â€” only needed when calling methods that use
them: ``CocoDetection.__getitem__`` needs PIL, and ``synchronize_between_processes``
needs torch.distributed).
é    )ÚannotationsN)Údefaultdicté   )Úmask)ÚCOCOÚCOCOevalÚParamsc                  ó2   — e Zd ZdZdd„Zd„ Zd„ Zd„ Zd„ ZdS )	ÚCocoDetectionu÷  A COCO-format detection dataset compatible with ``torch.utils.data.DataLoader``.

    Drop-in replacement for ``torchvision.datasets.CocoDetection`` backed by
    hotcoco instead of pycocotools. Does not inherit from any torch class â€”
    duck-typing is sufficient for ``DataLoader``.

    Parameters
    ----------
    root : str
        Root directory of images.
    ann_file : str
        Path to COCO-format annotation JSON file.
    transform : callable, optional
        Transform applied to the PIL image.
    target_transform : callable, optional
        Transform applied to the list of annotation dicts.
    transforms : callable, optional
        Joint transform applied to ``(image, target)``. Applied *after*
        individual transforms.
    Nc                ó¼   — || _         t          |¦  «        | _        t          | j                             ¦   «         ¦  «        | _        || _        || _        || _        d S ©N)	Úrootr   ÚcocoÚsortedÚget_img_idsÚidsÚ	transformÚtarget_transformÚ
transforms)Úselfr   Úann_filer   r   r   s         úd/root/private/project/pallet_detection_v2/.venv/lib/python3.11/site-packages/hotcoco/integrations.pyÚ__init__zCocoDetection.__init__-   sM   € ØˆŒ	Ý˜‘N”NˆŒ	Ý˜$œ)×/Ò/Ñ1Ô1Ñ2Ô2ˆŒØ"ˆŒØ 0ˆÔØ$ˆŒˆˆó    c                ó6  — 	 ddl m} n# t          $ r t          d¦  «        d ‚w xY w| j                             |g¦  «        d         }t
          j                             | j        |d         ¦  «        }| 	                    |¦  «         
                    d¦  «        S )Nr   )ÚImagezIPillow is required for CocoDetection. Install it with: pip install PillowÚ	file_nameÚRGB)ÚPILr   ÚImportErrorr   Ú	load_imgsÚosÚpathÚjoinr   ÚopenÚconvert)r   Úimg_idr   Úimg_infor#   s        r   Ú_load_imagezCocoDetection._load_image5   s¦   € ð	uØ!Ð!Ð!Ð!Ð!Ð!Ð!øÝð 	uð 	uð 	uÝÐiÑjÔjÐptÐtð	uøøøà”9×&Ò&¨ xÑ0Ô0°Ô3ˆÝŒw�|Š|˜DœI x°Ô'<Ñ=Ô=ˆØ�zŠz˜$ÑÔ×'Ò'¨Ñ.Ô.Ð.s   ‚	 ‰$c                ój  — | j         |         }| j                             |g¬¦  «        }| j                             |¦  «        }|                      |¦  «        }| j        �|                      |¦  «        }| j        �|                      |¦  «        }| j        �|                      ||¦  «        \  }}||fS )N)Úimg_ids)r   r   Úget_ann_idsÚ	load_annsr)   r   r   r   )r   Úindexr'   Úann_idsÚtargetÚimages         r   Ú__getitem__zCocoDetection.__getitem__>   s­   € Ø”˜%”ˆØ”)×'Ò'°°Ð'Ñ9Ô9ˆØ”×$Ò$ WÑ-Ô-ˆØ× Ò  Ñ(Ô(ˆàŒ>Ð%Ø—N’N 5Ñ)Ô)ˆEØÔ Ð,Ø×*Ò*¨6Ñ2Ô2ˆFØŒ?Ð&Ø ŸOšO¨E°6Ñ:Ô:‰MˆE�6à�fˆ}Ðr   c                ó*   — t          | j        ¦  «        S r   )Úlenr   ©r   s    r   Ú__len__zCocoDetection.__len__M   s   € Ý�4”8‰}Œ}Ðr   c                óP   — | j         j        › d| j        ›dt          | ¦  «        › d�S )Nz(root=z, num_images=ú))Ú	__class__Ú__name__r   r4   r5   s    r   Ú__repr__zCocoDetection.__repr__P   s.   € Ø”.Ô)ÐWÐW°´ÐWÐWÍ3ÈtÉ9Ì9ÐWÐWÐWÐWr   )NNN)	r:   Ú
__module__Ú__qualname__Ú__doc__r   r)   r2   r6   r;   © r   r   r   r      ss   € € € € € ðð ð*%ð %ð %ð %ð/ð /ð /ðð ð ðð ð ðXð Xð Xð Xð Xr   r   c                  ó6   — e Zd ZdZd„ Zd„ Zd„ Zd„ Zd„ Zd„ Z	dS )	ÚCocoEvaluatorab  Distributed COCO evaluator for PyTorch training loops.

    Drop-in replacement for the ``CocoEvaluator`` in torchvision's detection
    reference scripts. Wraps hotcoco's ``COCOeval`` with a tensor-friendly
    ``update()`` interface and optional distributed synchronization.

    Parameters
    ----------
    coco_gt : COCO
        Ground-truth COCO object.
    iou_types : list[str]
        IoU types to evaluate, such as ``["bbox"]`` or ``["bbox", "segm"]``.

    Example
    -------
    ::

        evaluator = CocoEvaluator(coco_gt, ["bbox"])
        for images, targets in data_loader:
            outputs = model(images)
            predictions = {t["image_id"]: o for t, o in zip(targets, outputs)}
            evaluator.update(predictions)
        evaluator.synchronize_between_processes()
        evaluator.accumulate()
        evaluator.summarize()
    c                ó°   — t          |t          ¦  «        r|g}|| _        || _        i | _        t          t          ¦  «        | _        |D ]}d | j        |<   Œd S r   )Ú
isinstanceÚstrÚcoco_gtÚ	iou_typesÚ	coco_evalr   ÚlistÚresults)r   rE   rF   Úiou_types       r   r   zCocoEvaluator.__init__p   sg   € Ý�i¥Ñ%Ô%ð 	$Ø"˜ˆIØˆŒØ"ˆŒØˆŒÝ"¥4Ñ(Ô(ˆŒà!ð 	,ð 	,ˆHØ'+ˆDŒN˜8Ñ$Ð$ð	,ð 	,r   c                óz   — | j         D ]2}t          ||¦  «        }| j        |                              |¦  «         Œ3dS )a‘  Accumulate predictions from one batch.

        Parameters
        ----------
        predictions : dict[int, dict]
            Mapping from image ID to a dict with keys depending on iou_type:

            - ``"bbox"``: ``{"boxes": Tensor(N,4), "scores": Tensor(N), "labels": Tensor(N)}``
              Boxes are in XYXY format; converted to XYWH internally.
            - ``"segm"``: ``{"masks": Tensor(N,1,H,W), "scores": Tensor(N), "labels": Tensor(N)}``
              Masks are thresholded at 0.5 and RLE-encoded internally.
            - ``"keypoints"``: ``{"keypoints": Tensor(N,K,3), "scores": Tensor(N),
              "labels": Tensor(N)}``
        N)rF   Ú_prepare_for_cocorI   Úextend)r   ÚpredictionsrJ   rI   s       r   ÚupdatezCocoEvaluator.update{   sL   € ð œð 	3ð 	3ˆHÝ'¨°XÑ>Ô>ˆGØŒL˜Ô"×)Ò)¨'Ñ2Ô2Ð2Ð2ð	3ð 	3r   c                ó  ‡‡— 	 ddl m} |                     ¦   «         sdS n# t          $ r Y dS w xY wddl}ddlŠ| j        D �]É}|                     | j        |         ¦  «        }‰ 	                    t          |¦  «        ‰j        d¬¦  «        }‰ 	                    t          |¦  «        g‰j        ¬¦  «        }|                     ¦   «         }ˆfd„t          |¦  «        D ¦   «         }|                     ||¦  «         t#          t%          d„ |D ¦   «         ¦  «        ¦  «        Š‰                     ‰‰j        ¬¦  «        }	||	dt          |¦  «        …<   ˆˆfd„t          |¦  «        D ¦   «         }
|                     |
|	¦  «         g }t)          |
|¦  «        D ]h\  }}t+          |d|                     ¦   «         …                              ¦   «         ¦  «        }|                     |                     |¦  «        ¦  «         Œi|| j        |<   �ŒËdS )	zwGather results across all distributed processes.

        No-op when ``torch.distributed`` is not initialized.
        r   NÚcpu)ÚdtypeÚdevice©rR   c                óJ   •— g | ]}‰                      d g‰j        ¬¦  «        ‘Œ S )r   rT   )ÚtensorÚlong)Ú.0Ú_Útorchs     €r   ú
<listcomp>z?CocoEvaluator.synchronize_between_processes.<locals>.<listcomp>¥   s-   ø€ ÐTÐTÐT¸Q�U—\’\ 1 #¨U¬Z�\Ñ8Ô8ÐTÐTÐTr   c              3  ó>   K  — | ]}|                      ¦   «         V — Œd S r   )Úitem)rX   Úss     r   ú	<genexpr>z>CocoEvaluator.synchronize_between_processes.<locals>.<genexpr>§   s*   è è € Ð7Ð7¨A˜qŸvšv™xœxÐ7Ð7Ð7Ð7Ð7Ð7r   c                óH   •— g | ]}‰                      ‰‰j        ¬ ¦  «        ‘ŒS )rT   )ÚzerosÚuint8)rX   rY   Úmax_sizerZ   s     €€r   r[   z?CocoEvaluator.synchronize_between_processes.<locals>.<listcomp>«   s+   ø€ Ð\Ð\Ð\ÀQ˜Ÿš H°E´K˜Ñ@Ô@Ð\Ð\Ð\r   )Útorch.distributedÚdistributedÚis_initializedr    ÚpicklerZ   rF   ÚdumpsrI   rV   rH   rb   r4   rW   Úget_world_sizeÚrangeÚ
all_gatherÚintÚmaxra   ÚzipÚbytesr]   ÚtolistrM   Úloads)r   Údistrg   rJ   Ú
local_dataÚlocal_tensorÚ
local_sizeÚ
world_sizeÚsizesÚpaddedÚgatheredÚall_resultsÚbufÚsizeÚdatarc   rZ   s                  @@r   Úsynchronize_between_processesz+CocoEvaluator.synchronize_between_processesŽ   s2  øø€ ð
	Ø,Ð,Ð,Ð,Ð,Ð,à×&Ò&Ñ(Ô(ð Ø�ðøåð 	ð 	ð 	ØˆFˆFð	øøøð 	ˆˆˆàˆˆˆàœð 	1ñ 	1ˆHØŸš d¤l°8Ô&<Ñ=Ô=ˆJØ Ÿ<š<­¨ZÑ(8Ô(8ÀÄÐTY˜<ÑZÔZˆLØŸš¥s¨:¡¤Ð&7¸u¼z˜ÑJÔJˆJà×,Ò,Ñ.Ô.ˆJØTÐTÐTÐTÅ%È
ÑBSÔBSÐTÑTÔTˆEØ�OŠO˜E :Ñ.Ô.Ð.Ý�3Ð7Ð7°Ð7Ñ7Ô7Ñ7Ô7Ñ8Ô8ˆHà—[’[ °´�[Ñ=Ô=ˆFØ(4ˆFÐ$•S˜‘_”_Ð$Ñ%Ø\Ð\Ð\Ð\Ð\Í%ÐPZÑJ[ÔJ[Ð\Ñ\Ô\ˆHØ�OŠO˜H fÑ-Ô-Ð-àˆKÝ  ¨5Ñ1Ô1ð 7ð 7‘	��TÝ˜S  4§9¢9¡;¤; Ô/×6Ò6Ñ8Ô8Ñ9Ô9�Ø×"Ò" 6§<¢<°Ñ#5Ô#5Ñ6Ô6Ð6Ð6Ø%0ˆDŒL˜Ñ"Ñ"ð'	1ð 	1s   „! ¡
/®/c                ó¦  — | j         D ]È}| j        |         }|st          d|›�¦  «         Œ$| j                             |¦  «        }t          | j        ||¦  «        }t          |¬¦  «        }t          | j                             ¦   «         ¦  «        |_	        ||_
        |                     ¦   «          |                     ¦   «          || j        |<   ŒÉdS )z6Create COCOeval objects and run evaluate + accumulate.z!Warning: no results for iou_type=)rJ   N)rF   rI   ÚprintrE   Úload_resr   r	   r   r   r+   ÚparamsÚevaluateÚ
accumulaterG   )r   rJ   rI   Úcoco_dtrG   r‚   s         r   r„   zCocoEvaluator.accumulate´   sÝ   € àœð 	1ð 	1ˆHØ”l 8Ô,ˆGØð ÝÐF¸(ÐFÐFÑGÔGÐGØà”l×+Ò+¨GÑ4Ô4ˆGÝ  ¤¨w¸ÑAÔAˆIõ  XÐ.Ñ.Ô.ˆFÝ# D¤L×$<Ò$<Ñ$>Ô$>Ñ?Ô?ˆFŒNØ%ˆIÔà×ÒÑ Ô Ð Ø× Ò Ñ"Ô"Ð"Ø'0ˆDŒN˜8Ñ$Ð$ð#	1ð 	1r   c                óÄ   — | j         D ]W}| j                             |¦  «        }|€t          d|›�¦  «         Œ1t          d|› �¦  «         |                     ¦   «          ŒXdS )z+Print evaluation metrics for each iou_type.Nz$Warning: no evaluation for iou_type=zIoU metric: )rF   rG   Úgetr€   Ú	summarize)r   rJ   Úevs      r   rˆ   zCocoEvaluator.summarizeÉ   sw   € àœð 	ð 	ˆHØ”×#Ò# HÑ-Ô-ˆBØˆzÝÐI¸XÐIÐIÑJÔJÐJØÝÐ+ Ð+Ð+Ñ,Ô,Ð,Ø�LŠL‰NŒNˆNˆNð	ð 	r   c                ó„   — i }| j         D ]5}| j                             |¦  «        }|�|                     ¦   «         ||<   Œ6|S )z7Return metrics as ``{iou_type: {metric_name: value}}``.)rF   rG   r‡   Úget_results)r   rI   rJ   r‰   s       r   r‹   zCocoEvaluator.get_resultsÓ   sL   € àˆØœð 	5ð 	5ˆHØ”×#Ò# HÑ-Ô-ˆBØˆ~Ø$&§N¢NÑ$4Ô$4�˜Ñ!øØˆr   N)
r:   r<   r=   r>   r   rO   r~   r„   rˆ   r‹   r?   r   r   rA   rA   T   sy   € € € € € ðð ð6	,ð 	,ð 	,ð3ð 3ð 3ð&$1ð $1ð $1ðL1ð 1ð 1ð*ð ð ðð ð ð ð r   rA   c                ó  — g }|                       ¦   «         D �]k\  }}t          |t          ¦  «        r|}nt          |¦  «        }t          |                     dg ¦  «        ¦  «        dk    rŒU|d                              ¦   «         }|d                              ¦   «         }|dk    ri|d         }|                     ¦   «         }t          |¦  «        D ]<\  }	}
|
\  }}}}|                     |||	         ||||z
  ||z
  g||	         dœ¦  «         Œ=Œø|dk    �r|d         }|                     ¦   «          	                    ¦   «         }t          |j        d         ¦  «        D ]¸}	||	df         }|d	k                         d
¦  «        }t          j        t          j        |¦  «        ¦  «        }|d         t          |d         t"          ¦  «        r|d                              d¦  «        n|d         dœ}|                     |||	         |||	         dœ¦  «         Œ¹�Œ|dk    rc|d         }|                     ¦   «         }t          |¦  «        D ]7\  }	}d„ |D ¦   «         }|                     |||	         |||	         dœ¦  «         Œ8�Œm|S )z;Convert tensor predictions to COCO-format annotation dicts.Úscoresr   ÚlabelsÚbboxÚboxes)Úimage_idÚcategory_idr�   ÚscoreÚsegmÚmasksg      à?rb   r|   Úcountszutf-8)r|   r–   )r‘   r’   Úsegmentationr“   Ú	keypointsc                ó   — g | ]	}|D ]}|‘ŒŒ
S r?   r?   )rX   ÚpointÚvs      r   r[   z%_prepare_for_coco.<locals>.<listcomp>  s%   € Ð;Ð;Ð;˜e°UÐ;Ð;°˜Ð;Ð;Ð;Ð;r   )r‘   r’   r˜   r“   )ÚitemsrC   rl   r4   r‡   rp   Ú	enumerateÚappendrQ   Únumpyrj   ÚshapeÚastypeÚ
mask_utilsÚencodeÚ_npÚasfortranarrayro   Údecode)rN   rJ   Úcoco_resultsr'   Ú
predictionÚoriginal_idr�   rŽ   r�   ÚiÚboxÚx1Úy1Úx2Úy2r•   r   ÚbinaryÚrleÚrle_segr˜   ÚkptsÚflats                          r   rL   rL   Ý   sâ  € à€LØ)×/Ò/Ñ1Ô1ð 4ñ 4Ñˆ�
Ý�f�cÑ"Ô"ð 	&Ø ˆKˆKå˜f™+œ+ˆKåˆz�~Š~˜h¨Ñ+Ô+Ñ,Ô,°Ò1Ð1Øà˜HÔ%×,Ò,Ñ.Ô.ˆØ˜HÔ%×,Ò,Ñ.Ô.ˆà�vÒÐØ˜wÔ'ˆEà—L’L‘N”NˆEÝ# EÑ*Ô*ð 	ð 	‘��3Ø!$‘��B˜˜BØ×#Ò#à$/Ø'-¨a¤yØ!# R¨¨b©°"°r±'Ð :Ø!'¨¤ð	ð ñô ð ð ð	ð ˜ÒÑØ˜wÔ'ˆEà—I’I‘K”K×%Ò%Ñ'Ô'ˆEÝ˜5œ; qœ>Ñ*Ô*ð ð �Ø˜Q ˜T”{�Ø š*×,Ò,¨WÑ5Ô5�Ý Ô'­Ô(:¸6Ñ(BÔ(BÑCÔC�ð   œKÝ?IÈ#ÈhÌ-ÕY^Ñ?_Ô?_Ðr˜c (œm×2Ò2°7Ñ;Ô;Ð;ÐehÐiqÔerðð �ð ×#Ò#Ø!,¸VÀA¼YÐX_ÐjpÐqrÔjsÐtÐtñô ð ð ñð ˜Ò$Ð$Ø" ;Ô/ˆIà!×(Ò(Ñ*Ô*ˆIÝ$ YÑ/Ô/ð ð ‘��4Ø;Ð; tÐ;Ñ;Ô;�Ø×#Ò#Ø!,¸VÀA¼YÐUYÐdjÐklÔdmÐnÐnñô ð ð ùð Ðr   )r>   Ú
__future__r   r"   Úcollectionsr   rŸ   r¤   Ú r   r¢   Úhotcocor   r   r	   r   rA   rL   r?   r   r   ú<module>r¹      sõ   ðð	ð 	ð #Ð "Ð "Ð "Ð "Ð "à 	€	€	€	Ø #Ð #Ð #Ð #Ð #Ð #à Ð Ð Ð à  Ð  Ð  Ð  Ð  Ð  Ø +Ð +Ð +Ð +Ð +Ð +Ð +Ð +Ð +Ð +ð:Xð :Xð :Xð :Xð :Xñ :Xô :Xð :XðzFð Fð Fð Fð Fñ Fô Fð FðR9ð 9ð 9ð 9ð 9r   