§
    à©ºje  ã                  ó�  — d 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
mZmZmZmZmZmZ ddlmZ ddlmZmZmZmZ d	d
d	ddd	d	dœdId„Zdd
d	ddd	d	dœdJd"„Zd	d#dd
d	ddd	d	d$œ	dKd(„Zd	d	d	d	d#d
d	ddd	d	d)œd*„Zd+d	d	d	ddd	d	d,œdLd4„Zd5ddd	d	d6œdMd7„Zd5d8ddd	d	d9œdNd<„Zddd	d	d=œdOd?„Zd#dddd	d	d@œdPdC„Z ddd	d	d=œdQdE„Z!d5ddd	d	dFœdRdH„Z"d	S )Sz—Public plot functions: pr_curve, confusion_matrix, top_confusions, per_category_ap,
tide_errors, reliability_diagram, comparison_bar, category_deltas.
é    )Úannotations)ÚPath)ÚAnyé   )	ÚTIDE_ERROR_ORDERÚ_annotate_f1_peakÚ_configure_axesÚ_headroom_for_bar_labelsÚ_import_mplÚ_mask_invalid_precÚ_new_figureÚ_save_and_returnÚ_top_confusion_keep)ÚPlotData)ÚCHROMEÚ	_build_rcÚ
_get_themeÚeval_colorsNÚallÚ	cyanotypeF©Úiou_thrsÚarea_rngÚmax_detÚthemeÚ
paper_modeÚaxÚ	save_pathr   úlist[float] | Noner   Ústrr   ú
int | Noner   r   Úboolr   ústr | Path | NoneÚreturnÚtuplec          
     óâ  — t          ¦   «         \  }}	}	t          j        | ¦  «        }
|
                     |¦  «        }|
                     |¦  «        }|€.t          t          t          |
j        ¦  «        ¦  «        ¦  «        n|
 	                    |¦  «        }| 
                    t          ||¦  «        ¦  «        5  t          d|d¬¦  «        \  }}t          |¦  «        }t          |¦  «        D ]x\  }}|
                     |||¦  «        }|dk    rdnd}|                     |
j        ||d|
j        |         d	›�¬
¦  «        \  }|dk    rt#          ||
j        |||dk    ¬¦  «         Œy|                     ddddd¬¦  «         |dk    r|                     ddd¬¦  «         n|                     dd¬¦  «         t)          |ddd¬¦  «         ddd¦  «         n# 1 swxY w Y   t+          |||¦  «        S )uE  Plot mean precision-recall across all IoU thresholds.

    One line per IoU threshold, precision averaged across all categories. The
    primary line (lowest IoU) is annotated at its F1 peak, and filled under only
    when the plot has at most two curves â€” a fill runs to zero, so under the
    topmost curve of a wider sweep it tints every other curve's area too.

    Parameters
    ----------
    coco_eval : COCOeval
        Must have ``run()`` called first.
    iou_thrs : list[float], optional
        IoU thresholds to include. Default: all thresholds in params. Raises
        ``ValueError`` if a threshold is not in this run's grid.
    area_rng : str
        Area range label. Default ``"all"``.
    max_det : int, optional
        Max detections. Default: last entry in ``params.max_dets``.
    theme : str
        ``"cyanotype"`` (default) or ``"cyanotype-dark"``.
    paper_mode : bool
        White figure and axes background for PDF/LaTeX inclusion.
    ax : matplotlib.axes.Axes, optional
    save_path : str or Path, optional

    Returns
    -------
    (Figure, Axes)
    N©é   r(   Ú
compressed©Úlayoutr   é   r   úIoU=ú.2f©Ú	linewidthÚlabel)Úfill©r   r   ÚequalÚRecallÚ	Precision©ÚxlimÚylimÚaspectÚxlabelÚylabelé   é	   zcenter left)gR¸…ëQð?ç      à?)ÚfontsizeÚlocÚbbox_to_anchorú
lower left©r@   rA   úPrecision-Recallzmean over categoriesÚy©ÚsubtitleÚ
value_axis)r   r   Úfrom_coco_evalÚarea_idxÚmax_det_idxÚlistÚrangeÚlenÚiou_thresholdsÚiou_indicesÚ
rc_contextr   r   Ú	enumerateÚmean_precisionÚplotÚ
recall_ptsr   ÚsetÚlegendr	   r   )Ú	coco_evalr   r   r   r   r   r   r   ÚmplÚ_ÚdataÚa_idxÚm_idxÚ	t_indicesÚfigÚn_curvesÚline_idxÚt_idxÚprecÚlwÚlines                        úb/root/private/project/pallet_detection_v2/.venv/lib/python3.11/site-packages/hotcoco/plot/plots.pyÚpr_curve_iou_sweeprh      sO  € õP ‘”�I€CˆˆAÝÔ" 9Ñ-Ô-€DØ�MŠM˜(Ñ#Ô#€EØ×Ò˜WÑ%Ô%€Eà9AÐ9I••U�3˜tÔ2Ñ3Ô3Ñ4Ô4Ñ5Ô5Ð5Èt×O_ÒO_Ð`hÑOiÔOi€Ià	�Š�	 %¨Ñ4Ô4Ñ	5Ô	5ð að aÝ˜f b°Ð>Ñ>Ô>‰ˆˆRå�y‘>”>ˆÝ(¨Ñ3Ô3ð 	Wð 	W‰OˆH�eØ×&Ò& u¨e°UÑ;Ô;ˆDØ !’m�m��¨ˆBØ—g’g˜dœo¨t¸rÐIpÐPTÔPcÐdiÔPjÐIpÐIpÐIp�gÑqÔq‰GˆTØ˜1Š}ˆ}Ý! " d¤o°t¸TÈÐTUÊÐVÑVÔVÐVøà
�Š�F °ÀÐQ\ˆÑ]Ô]Ð]ð �aŠ<ˆ<Ø�IŠI˜q mÀKˆIÑPÔPÐPÐPà�IŠI˜q lˆIÑ3Ô3Ð3Ý˜Ð.Ð9OÐ\_Ð`Ñ`Ô`Ð`ð'að að añ aô að að að að að að aøøøð að að að aõ( ˜C  YÑ/Ô/Ð/s   Â9DGÇGÇGr?   ©Úiou_thrr   r   r   r   r   r   Úcat_idÚintrj   Úfloatc               óR  — t          ¦   «         \  }	}
}
t          j        | ¦  «        }|                     |¦  «        }|                     |¦  «        }||j        vrt          d|› d�¦  «        ‚|j                             |¦  «        }|                     |¦  «        }t          |j
        |dd…|||f         ¦  «        }|j                             |t          |¦  «        ¦  «        }|	                     t          ||¦  «        ¦  «        5  t!          d|d¬¦  «        \  }}|                     |j        |d¬¦  «        \  }t'          ||j        ||¦  «         |                     d	d	d
dd¬¦  «         t+          |d|d¬¦  «         ddd¦  «         n# 1 swxY w Y   t-          |||¦  «        S )aÁ  Plot the precision-recall curve for a single category.

    Parameters
    ----------
    coco_eval : COCOeval
        Must have ``run()`` called first.
    cat_id : int
        Category ID to plot.
    iou_thr : float
        IoU threshold. Default 0.50.
    area_rng : str
        Area range label. Default ``"all"``.
    max_det : int, optional
        Max detections. Default: last entry in ``params.max_dets``.
    theme : str
        ``"cyanotype"`` (default) or ``"cyanotype-dark"``.
    paper_mode : bool
        White figure and axes background for PDF/LaTeX inclusion.
    ax : matplotlib.axes.Axes, optional
    save_path : str or Path, optional

    Returns
    -------
    (Figure, Axes)
    zcat_id z not in params.cat_idsNr'   r)   r*   r,   )r0   r3   r4   r5   r6   r7   rE   rF   rG   )r   r   rJ   rK   rL   Úcat_idsÚ
ValueErrorÚindexÚnearest_iou_idxr   Ú	precisionÚ	cat_namesÚgetr    rR   r   r   rU   rV   r   rW   r	   r   )rY   rk   rj   r   r   r   r   r   r   rZ   r[   r\   r]   r^   Úk_idxrc   rd   Úcat_namer`   rf   s                       rg   Úpr_curve_by_categoryrx   _   së  € õL ‘”�I€CˆˆAÝÔ" 9Ñ-Ô-€DØ�MŠM˜(Ñ#Ô#€EØ×Ò˜WÑ%Ô%€Eà�T”\Ð!Ð!ÝÐA 6ÐAÐAÐAÑBÔBÐBàŒL×Ò˜vÑ&Ô&€EØ× Ò  Ñ)Ô)€EÝ˜dœn¨U°A°A°A°u¸eÀUÐ-JÔKÑLÔL€DØŒ~×!Ò! &­#¨f©+¬+Ñ6Ô6€Hà	�Š�	 %¨Ñ4Ô4Ñ	5Ô	5ð Sð SÝ˜f b°Ð>Ñ>Ô>‰ˆˆRØ—'’'˜$œ/¨4¸1�'Ñ=Ô=‰ˆÝ˜"˜dœo¨t°TÑ:Ô:Ð:à
�Š�F °ÀÐQ\ˆÑ]Ô]Ð]Ý˜Ð.¸ÈcÐRÑRÔRÐRðSð Sð Sñ Sô Sð Sð Sð Sð Sð Sð Søøøð Sð Sð Sð Sõ ˜C  YÑ/Ô/Ð/s   ÄA9FÆFÆFé
   ©	ro   Útop_nrj   r   r   r   r   r   r   ro   úlist[int] | Noner{   c       	        óà  — ddl }
t          ¦   «         \  }}}t          j        | ¦  «        }|                     |¦  «        }|                     |¦  «        }|                     |¦  «        }|€°g }t          |j        ¦  «        D ]m\  }}|j	        |dd…|||f         }||dk             }| 
                    |t          |¦  «        r"t          |
                     |¦  «        ¦  «        ndf¦  «         Œn|                     d„ d¬¦  «         d„ |d|…         D ¦   «         }d„ t          |j        ¦  «        D ¦   «         }|                     t!          ||¦  «        ¦  «        5  t#          d	|d
¬¦  «        \  }}|D ]}|                     |¦  «        }|€Œt'          |j	        |dd…|||f         ¦  «        }|                     |j        |d|j                             |t/          |¦  «        ¦  «        ¬¦  «         Œ€|                     ddddd¬¦  «         |                     dd¬¦  «         t5          |dd|j        |         d›�d¬¦  «         ddd¦  «         n# 1 swxY w Y   t9          |||	¦  «        S )a^  Plot precision-recall curves for multiple categories.

    Parameters
    ----------
    coco_eval : COCOeval
        Must have ``run()`` called first.
    cat_ids : list[int], optional
        Categories to plot. If omitted, the top ``top_n`` by AP are selected.
    top_n : int
        Number of top categories when ``cat_ids`` is not given. Default 10.
    iou_thr : float
        IoU threshold. Default 0.50.
    area_rng : str
        Area range label. Default ``"all"``.
    max_det : int, optional
        Max detections. Default: last entry in ``params.max_dets``.
    theme : str
        ``"cyanotype"`` (default) or ``"cyanotype-dark"``.
    paper_mode : bool
        White figure and axes background for PDF/LaTeX inclusion.
    ax : matplotlib.axes.Axes, optional
    save_path : str or Path, optional

    Returns
    -------
    (Figure, Axes)
    r   Nç        c                ó   — | d         S ©Nr   © ©Úxs    rg   ú<lambda>z pr_curve_top_n.<locals>.<lambda>Ò   s
   €  1 Q¤4€ ó    T©ÚkeyÚreversec                ó   — g | ]\  }}|‘ŒS r�   r�   )Ú.0Úcidr[   s      rg   ú
<listcomp>z"pr_curve_top_n.<locals>.<listcomp>Ó   s   € Ð5Ð5Ð5™6˜3 �3Ð5Ð5Ð5r…   c                ó   — i | ]\  }}||“Œ	S r�   r�   )rŠ   Úkr‹   s      rg   ú
<dictcomp>z"pr_curve_top_n.<locals>.<dictcomp>Õ   s   € Ð@Ð@Ð@™f˜a �3˜Ð@Ð@Ð@r…   r'   r)   r*   g      ø?r/   r3   r4   r5   r6   r7   é   rC   rD   zPrecision-Recall by Categoryr-   r.   rF   rG   )Únumpyr   r   rJ   rK   rL   rr   rS   ro   rs   ÚappendrO   rm   ÚmeanÚsortrR   r   r   ru   r   rU   rV   rt   r    rW   rX   r	   rP   r   )rY   ro   r{   rj   r   r   r   r   r   r   ÚnprZ   r[   r\   r]   r^   rc   Úcat_apsÚk_ir‹   ÚpÚvalidÚcat_id_to_kr`   rv   rd   s                             rg   Úpr_curve_top_nr›   œ   sÔ  € ðP ÐÐÐå‘”�I€CˆˆAÝÔ" 9Ñ-Ô-€DØ�MŠM˜(Ñ#Ô#€EØ×Ò˜WÑ%Ô%€EØ× Ò  Ñ)Ô)€Eà€ØˆÝ! $¤,Ñ/Ô/ð 	Pð 	P‰HˆC�Ø”˜u a a a¨¨e°UÐ:Ô;ˆAØ�a˜1’f”IˆEØ�NŠN˜C½#¸e¹*¼*Ð!M¥ r§w¢w¨u¡~¤~Ñ!6Ô!6Ð!6È#ÐNÑOÔOÐOÐOØ�Š˜˜°ˆÑ6Ô6Ð6Ø5Ð5 W¨V¨e¨V¤_Ð5Ñ5Ô5ˆà@Ð@­	°$´,Ñ(?Ô(?Ð@Ñ@Ô@€Kà	�Š�	 %¨Ñ4Ô4Ñ	5Ô	5ð 
ð 
Ý˜f b°Ð>Ñ>Ô>‰ˆˆRàð 	cð 	cˆCØ—O’O CÑ(Ô(ˆEØˆ}ØÝ% d¤n°U¸A¸A¸A¸uÀeÈUÐ5RÔ&SÑTÔTˆDØ�GŠG�D”O T°SÀÄ×@RÒ@RÐSVÕX[Ð\_ÑX`ÔX`Ñ@aÔ@aˆGÑbÔbÐbÐbà
�Š�F °ÀÐQ\ˆÑ]Ô]Ð]Ø
�	Š	˜1 ,ˆ	Ñ/Ô/Ð/ÝØÐ.Ð9`ÀÔ@SÐTYÔ@ZÐ9`Ð9`Ð9`Ðmpð	
ñ 	
ô 	
ð 	
ð
ð 
ð 
ñ 
ô 
ð 
ð 
ð 
ð 
ð 
ð 
øøøð 
ð 
ð 
ð 
õ ˜C  YÑ/Ô/Ð/s   ÅC+IÉIÉI)r   rk   ro   rj   r{   r   r   r   r   r   r   c               óª   — |�|nd}|�t          | ||||||	|
|¬¦	  «	        S |€|�t          | |||||||	|
|¬¦
  «
        S t          | |||||	|
|¬¦  «        S )a€  Plot a precision-recall curve, dispatching on the arguments given.

    ``cat_id=`` plots a single category (``pr_curve_by_category``);
    ``cat_ids=`` or ``iou_thr=`` plots one curve per category
    (``pr_curve_top_n``); with neither, one curve per IoU threshold
    (``pr_curve_iou_sweep``). The named functions take the same keyword
    arguments and can be called directly.
    Nr?   ri   rz   r   )rx   r›   rh   )rY   r   rk   ro   rj   r{   r   r   r   r   r   r   Úresolved_iou_thrs                rg   Úpr_curverž   é   s¾   € ð2 #*Ð"5�w�w¸3ÐØÐÝ#ØØØ$ØØØØ!ØØð

ñ 

ô 

ð 
	
ð Ð˜gÐ1ÝØØØØ$ØØØØ!ØØð
ñ 
ô 
ð 	
õ ØØØØØØØØð	ñ 	ô 	ð 	r…   T)Ú	normalizer{   Úgroup_byÚ
cat_groupsr   r   r   r   Úcm_dictúdict[str, Any]rŸ   r    ú
str | Noner¡   údict[str, list[str]] | Nonec               ó°	  ‡.‡/— ddl }	t          ¦   «         \  }
}}|	                     | d         t          ¬¦  «        }|	                     | d         t          ¬¦  «        }t	          | d         ¦  «        }t          |¦  «        }|dk    �r±|��®d„ t          |¦  «        D ¦   «         }t          |                     ¦   «         ¦  «        }t          |¦  «        }|	 	                    |d	z   |d	z   ft          ¬¦  «        }i }t          |¦  «        D ]\  }}||         D ]}||v r|||<   ŒŒt          |¦  «        D ]~\  }}| 
                    |¦  «        }|€Œt          |¦  «        D ]7\  }}| 
                    |¦  «        }|€Œ|||fxx         |||f         z  cc<   Œ8|||fxx         |||f         z  cc<   Œt          |¦  «        D ]6\  }}| 
                    |¦  «        }|�|||fxx         |||f         z  cc<   Œ7|||f         |||f<   |r&|                     d	d
¬¦  «        }d	||dk    <   ||z  }n|}|dgz   Š/|}n|dgz   Š/|r|n|}|}t          |||¦  «        } | �*||	                     | | ¦  «                 }ˆ/fd„| D ¦   «         Š/t          ‰/¦  «        }!t          t          dd|!z  ¦  «        d¦  «        }"|rdnd}#t!          ||¦  «        }$|
                     |$¦  «        5  t%          |"|"f|d¬¦  «        \  }%}|                     |dd|#d¬¦  «        Š.|rdnd	}&d/ˆ.fd„}'t)          |!¦  «        D ]�}t)          |!¦  «        D ]o}|||f         }(|(|&k     rŒ |'|(¦  «        })|r|(d›nt+          |(¦  «        › }*|                     |||*dd|)t          dt          dd |!z  ¦  «        ¦  «        ¬!¦  «         ŒpŒ‚|                     t)          |!¦  «        ¦  «         |                     t)          |!¦  «        ¦  «         |                     ‰/d"d#¬$¦  «         |                     ‰/¦  «         |                     d%¦  «         |                     d&¦  «         |                     d'¦  «         t=          |d(d¬)¦  «         dd*lm }+  |+|¦  «        },|, !                    d#d+d,¬-¦  «        }-|% "                    ‰.|-¬.¦  «         ddd¦  «         n# 1 swxY w Y   tG          |%||¦  «        S )0aû  Plot a confusion matrix heatmap.

    Parameters
    ----------
    cm_dict : dict
        Output of ``coco_eval.confusion_matrix()``.
    normalize : bool
        Use row-normalized values (default True).
    top_n : int, optional
        Show only the top N categories by off-diagonal confusion mass.
        Auto-set to 25 when there are more than 30 categories and no explicit value is given.
    group_by : str, optional
        ``"supercategory"`` to aggregate into COCO supercategory groups.
        Requires ``cat_groups`` mapping.
    cat_groups : dict[str, list[str]], optional
        Mapping of group name to list of category names for ``group_by``.
    theme : str
        ``"cyanotype"`` (default) or ``"cyanotype-dark"``.
    paper_mode : bool
        White figure and axes background for PDF/LaTeX inclusion.
    ax : matplotlib.axes.Axes, optional
        Existing axes to draw on.
    save_path : str or Path, optional
        Save figure to this path.

    Returns
    -------
    (Figure, Axes)
    r   NÚmatrix©ÚdtypeÚ
normalizedrt   Úsupercategoryc                ó   — i | ]\  }}||“Œ	S r�   r�   )rŠ   ÚiÚns      rg   r�   z$confusion_matrix.<locals>.<dictcomp>\  s   € Ð=Ð=Ð=¡  1�q˜!Ð=Ð=Ð=r…   r   T)ÚaxisÚkeepdimsÚBGc                ó    •— g | ]
}‰|         ‘ŒS r�   r�   )rŠ   r­   Úlabelss     €rg   rŒ   z$confusion_matrix.<locals>.<listcomp>‹  s   ø€ Ð*Ð*Ð* �&˜”)Ð*Ð*Ð*r…   r(   çffffffÖ?é   g      ð?r*   r4   Únone)r:   ÚvminÚvmaxÚinterpolationg{®Gáz„?Úvaluerm   r$   r    c                ó  •— ‰                      ‰                     | ¦  «        ¦  «        \  }}}}d„ |||fD ¦   «         }d|d         z  d|d         z  z   d|d         z  z   }|dk     rt          d	         nt          d
         S )a"  Pick the ink that survives on this cell.

            Driven by the cell's own luminance, not the theme: on a light theme
            the high cells are dark navy and on a dark theme they are bright
            steel, so a fixed "white for high values" is wrong in one of them.
            c                ó:   — g | ]}|d k    r|dz  n
|dz   dz  dz  ‘ŒS )g#Æ‚¤?g×£p=
×)@g)\�Âõ(¬?gáz®Gáð?g333333@r�   ©rŠ   Úcs     rg   rŒ   z2confusion_matrix.<locals>._ink.<locals>.<listcomp>¡  s7   € ÐcÐcÐcÐTU  g¢ �A˜‘I�I°Q¸±YÀ%Ñ4GÈCÑ3OÐcÐcÐcr…   g¼–�z6Ë?r   g¥,Cëâæ?r   g]mÅþ²{²?r,   gÍÌÌÌÌÌÜ?Úplot_bgÚtext)ÚcmapÚnormr   )rº   ÚrÚgÚbr[   ÚchanÚlumÚims          €rg   Ú_inkzconfusion_matrix.<locals>._ink™  s‹   ø€ ð Ÿš §¢¨¡¤Ñ0Ô0‰JˆAˆq�!�QØcÐcÐZ[Ð]^Ð`aÐYbÐcÑcÔcˆDØ˜4 œ7Ñ" V¨d°1¬gÑ%5Ñ5¸ÀÀaÄÑ8HÑHˆCØ(+¨dª
¨
•6˜)Ô$Ð$½¸v¼ÐFr…   r.   Úcenteré   r>   éd   )ÚhaÚvaÚcolorr@   é-   Úright)ÚrotationrÍ   Ú	PredictedzGround truthFzConfusion Matrix©rI   )Úmake_axes_locatablez5%g{®Gáz´?)ÚsizeÚpad)Úcax)rº   rm   r$   r    )$r‘   r   Úasarrayrm   rM   rO   rS   ÚsortedÚkeysÚzerosru   Úsumr   Úix_ÚminÚmaxr   rR   r   ÚimshowrN   rl   rÀ   Ú
set_xticksÚ
set_yticksÚset_xticklabelsÚset_yticklabelsÚ
set_xlabelÚ
set_ylabelÚset_frame_onr	   Úmpl_toolkits.axes_grid1rÕ   Úappend_axesÚcolorbarr   )0r¢   rŸ   r{   r    r¡   r   r   r   r   r•   rZ   r[   Ú
raw_matrixÚnorm_matrixrt   Ún_catsÚname_to_idxÚgroup_namesÚn_groupsÚgroupedÚgroup_idx_mapÚgiÚgnameÚcnamer­   ÚinameÚjÚjnameÚgjÚrow_sumsr\   Ú	n_plottedÚkeepr®   rÖ   r¸   Úrcr`   ÚsuppressrÉ   ÚvalrÏ   rÀ   rÕ   ÚdividerrØ   rÈ   r³   s0                                                 @@rg   Úconfusion_matrixr  (  s  øø€ ðR ÐÐÐå‘”�I€CˆˆAà—’˜G HÔ-µU�Ñ;Ô;€JØ—*’*˜W \Ô2½%�*Ñ@Ô@€KÝ�W˜[Ô)Ñ*Ô*€IÝ�‰^Œ^€Fð �?Ò"Ñ" zÑ'=Ø=Ð=­	°)Ñ(<Ô(<Ð=Ñ=Ô=ˆÝ˜ZŸ_š_Ñ.Ô.Ñ/Ô/ˆÝ�{Ñ#Ô#ˆà—(’(˜H q™L¨(°Q©,Ð7½u�(ÑEÔEˆØˆÝ" ;Ñ/Ô/ð 	.ð 	.‰IˆB�Ø# EÔ*ð .ð .�Ø˜KÐ'Ð'Ø+-�M %Ñ(øð.õ " )Ñ,Ô,ð 		;ð 		;‰HˆAˆuØ×"Ò" 5Ñ)Ô)ˆBØˆzØÝ% iÑ0Ô0ð 4ð 4‘��5Ø"×&Ò& uÑ-Ô-�Ø�:ØØ˜˜B˜��” :¨a°¨dÔ#3Ñ3��‘�Ø�B˜�LÐ!Ð!Ô! Z°°6°	Ô%:Ñ:Ð!Ð!Ñ!Ð!Ý! )Ñ,Ô,ð 	?ð 	?‰HˆAˆuØ×"Ò" 5Ñ)Ô)ˆBØˆ~Ø˜ "˜Ð%Ð%Ô%¨°F¸A°IÔ)>Ñ>Ð%Ð%Ñ%øØ&0°¸°Ô&@ˆ�˜(Ð"Ñ#àð 	Ø—{’{¨°D�{Ñ9Ô9ˆHØ&'ˆH�X ’]Ñ#Ø˜XÑ%ˆDˆDàˆDà ˜vÑ%ˆØˆ	ˆ	à˜d˜VÑ#ˆØ'Ð7ˆ{ˆ{¨ZˆØˆ	õ
 ˜t Y°Ñ6Ô6€DØÐØ�B—F’F˜4 Ñ&Ô&Ô'ˆØ*Ð*Ð*Ð* TÐ*Ñ*Ô*ˆåˆF‰Œ€AÝ�s�1�d˜Q‘hÑÔ Ñ$Ô$€DØÐ%ˆ3ˆ3 €DÝ	�5˜*Ñ	%Ô	%€BØ	�Š˜Ñ	Ô	ð )"ð )"å˜t T˜l¨B°tÐ<Ñ<Ô<‰ˆˆRà�YŠY�t G°!¸$ÈfˆYÑUÔUˆà$Ð+�4�4¨!ˆð
	Gð 
	Gð 
	Gð 
	Gð 
	Gð 
	Gõ �q‘”ð 	mð 	mˆAÝ˜1‘X”Xð mð m�Ø˜1˜a˜4”j�Ø˜’>�>ØØ˜˜S™	œ	�Ø'0ÐC˜#�|�|�|½¸S¹¼°m�Ø—’˜˜1˜d x°HÀEÕTWÐXYÕ[^Ð_`ÐbeÐhiÑbiÑ[jÔ[jÑTkÔTk�ÑlÔlÐlÐlðmð 	�Š•e˜A‘h”hÑÔÐØ
�Š•e˜A‘h”hÑÔÐØ
×Ò˜6¨B°7ÐÑ;Ô;Ð;Ø
×Ò˜6Ñ"Ô"Ð"Ø
�Š�kÑ"Ô"Ð"Ø
�Š�nÑ%Ô%Ð%Ø
�Š˜ÑÔÐÝ˜Ð.¸4Ð@Ñ@Ô@Ð@Ø?Ð?Ð?Ð?Ð?Ð?à%Ð% bÑ)Ô)ˆØ×!Ò! '°¸$Ð!Ñ?Ô?ˆØ�Š�R˜SˆÑ!Ô!Ð!ðS)"ð )"ð )"ñ )"ô )"ð )"ð )"ð )"ð )"ð )"ð )"øøøð )"ð )"ð )"ð )"õV ˜C  YÑ/Ô/Ð/s   ËGR<Ò<S ÓS rµ   )r{   r   r   r   r   c          
     ó„  — ddl }t          ¦   «         \  }}}|                     | d         t          ¬¦  «        }	t	          | d         ¦  «        }
t          |
¦  «        }|
dgz   }g }t          |dz   ¦  «        D ]_}t          |dz   ¦  «        D ]J}||k    rŒ	t          |	||f         ¦  «        }|dk    r$|                     ||         ||         |f¦  «         ŒKŒ`|                     d„ d	¬
¦  «         |d|…         }| 	                    t          ||¦  «        ¦  «        5  |sct          d|¦  «        \  }}|                     dddddd|j        ¬¦  «         t          |dd¬¦  «         t          |||¦  «        cddd¦  «         S d„ |D ¦   «         }d„ |D ¦   «         }t          |¦  «        }t          dt!          dd|z  ¦  «        f|¦  «        \  }}|                     t          |¦  «        |d¬¦  «        }|                     |d„ |D ¦   «         dd¬¦  «         t'          ||¦  «         |                     t          |¦  «        ¦  «         |                     |¦  «         |                     ¦   «          |                     d¦  «         t          |ddd¬ ¦  «         ddd¦  «         n# 1 swxY w Y   t          |||¦  «        S )!a  Plot the top N most common misclassifications as horizontal bars.

    This is the go-to plot for large numbers of categories (>30) where a
    full confusion matrix heatmap is unreadable. Shows only off-diagonal
    mistakes: "ground truth X predicted as Y".

    Parameters
    ----------
    cm_dict : dict
        Output of ``coco_eval.confusion_matrix()``.
    top_n : int
        Number of top confusions to show. Default 20.
    theme : str
        ``"cyanotype"`` (default) or ``"cyanotype-dark"``.
    paper_mode : bool
        White figure and axes background for PDF/LaTeX inclusion.
    ax : matplotlib.axes.Axes, optional
        Existing axes to draw on.
    save_path : str or Path, optional
        Save figure to this path.

    Returns
    -------
    (Figure, Axes)
    r   Nr§   r¨   rt   r±   r   c                ó   — | d         S )Nr,   r�   r‚   s    rg   r„   z top_confusions.<locals>.<lambda>ó  s
   € ˜Q˜qœT€ r…   Tr†   )r�   é   r?   zNo confusions foundrÊ   é   )rÍ   rÎ   r@   Ú	transformzTop ConfusionsrÔ   c                ó$   — g | ]\  }}}|› d |› �‘ŒS )u    â†’ r�   )rŠ   ÚgtÚpredr[   s       rg   rŒ   z"top_confusions.<locals>.<listcomp>ý  s,   € ÐEÐEÐE±°°D¸!˜Ð+Ð+ TÐ+Ð+ÐEÐEÐEr…   c                ó   — g | ]\  }}}|‘Œ	S r�   r�   )rŠ   r[   r¾   s      rg   rŒ   z"top_confusions.<locals>.<listcomp>þ  s   € Ð)Ð)Ð)™˜˜1˜a�!Ð)Ð)Ð)r…   r�   r=   r´   çffffffæ?©Úheightc                ó,   — g | ]}t          |¦  «        ‘ŒS r�   )r    r½   s     rg   rŒ   z"top_confusions.<locals>.<listcomp>  s   € Ð":Ð":Ð":¨a¥3 q¡6¤6Ð":Ð":Ð":r…   r  ©r³   r@   ÚpaddingÚCountu   ground truth â†’ predictionrƒ   rG   )r‘   r   rÙ   rl   rM   rO   rN   r’   r”   rR   r   r   rÀ   Ú	transAxesr	   r   rà   ÚbarhÚ	bar_labelr
   rã   rå   Úinvert_yaxisræ   )r¢   r{   r   r   r   r   r•   rZ   r[   r§   rt   rî   r³   Úpairsr­   rø   Úcountr`   Ú
bar_labelsÚcountsÚnum_barsÚbarss                         rg   Útop_confusionsr  ¿  s[  € ðD ÐÐÐå‘”�I€CˆˆAà�ZŠZ˜ Ô)µˆZÑ5Ô5€FÝ�W˜[Ô)Ñ*Ô*€IÝ�‰^Œ^€FØ˜$˜Ñ€Fà€EÝ�6˜A‘:ÑÔð <ð <ˆÝ�v ‘zÑ"Ô"ð 	<ð 	<ˆAØ�AŠvˆvØÝ˜˜q !˜tœÑ%Ô%ˆEØ�qŠyˆyØ—’˜f Qœi¨°¬°EÐ:Ñ;Ô;Ð;øð	<ð 
‡J‚J�>�>¨4€JÑ0Ô0Ð0Ø�&�5�&ŒM€Eà	�Š�	 %¨Ñ4Ô4Ñ	5Ô	5ð ið iØð 	8Ý! &¨"Ñ-Ô-‰GˆC�Ø�GŠG�C˜Ð3¸ÀXÐXZÐfhÔfrˆGÑsÔsÐsÝ˜BÐ 0¸TÐBÑBÔBÐBÝ# C¨¨YÑ7Ô7ðið ið ið iñ iô ið ið ið FÐE¸uÐEÑEÔEˆ
Ø)Ð) 5Ð)Ñ)Ô)ˆÝ�z‘?”?ˆå˜q¥# a¨°©Ñ"9Ô"9Ð:¸BÑ?Ô?‰ˆˆRØ�wŠw•u˜X‘”¨°sˆwÑ;Ô;ˆØ
�Š�TÐ":Ð":°6Ð":Ñ":Ô":ÀQÐPQˆÑRÔRÐRÝ   VÑ,Ô,Ð,à
�Š•e˜H‘o”oÑ&Ô&Ð&Ø
×Ò˜:Ñ&Ô&Ð&Ø
�ŠÑÔÐØ
�Š�gÑÔÐÝ˜Ð,Ð7WÐdgÐhÑhÔhÐhð)ið ið iñ iô ið ið ið ið ið ið iøøøð ið ið ið iõ, ˜C  YÑ/Ô/Ð/s   ÄAJ&ÆDJ&Ê&J*Ê-J*rË   )r{   Úbottom_nr   r   r   r   Úresults_dictr  c          	     óš  ‡‡— t          ¦   «         \  }}}|                      di ¦  «        }	|	st          d¦  «        ‚t          |	                     ¦   «         d„ d¬¦  «        }
t          |
¦  «        ||z   k    r|
d|…         dgz   |
| d…         z   }
d„ |
D ¦   «         }d	„ |
D ¦   «         }t          |¦  «        }|                     t          ||¦  «        ¦  «        5  t          d
t          dd|z  ¦  «        f|¦  «        \  }}t          t          |j        d         ¦  «        ¦  «        d         Š|j                             dd¦  «        Šˆˆfd„|D ¦   «         }|                     t          |¦  «        |d|¬¦  «        }d„ t          ||¦  «        D ¦   «         }|                     ||dd¬¦  «         t#          ||¦  «         d„ t          ||¦  «        D ¦   «         }|rt%          |¦  «        t          |¦  «        z  nd}|                     |ddd|d›�¬¦  «         |                     t          |¦  «        ¦  «         |                     |¦  «         |                     ¦   «          |                     d¦  «         |                     d d!¬"¦  «         t3          |d#d$¬%¦  «         ddd¦  «         n# 1 swxY w Y   t5          |||¦  «        S )&ax  Plot per-category AP as horizontal bars.

    Parameters
    ----------
    results_dict : dict
        Output of ``coco_eval.results(per_class=True)``.
    top_n : int
        Number of top categories to show.
    bottom_n : int
        Number of bottom categories to show.
    theme : str
        ``"cyanotype"`` (default) or ``"cyanotype-dark"``.
    paper_mode : bool
        White figure and axes background for PDF/LaTeX inclusion.
    ax : matplotlib.axes.Axes, optional
        Existing axes to draw on.
    save_path : str or Path, optional
        Save figure to this path.

    Returns
    -------
    (Figure, Axes)
    Ú	per_classz0No per-class data. Call results(per_class=True).c                ó   — | d         S r€   r�   r‚   s    rg   r„   z!per_category_ap.<locals>.<lambda>6  s
   € °A°a´D€ r…   Tr†   N)ú...Nc                ó   — g | ]
}|d          ‘ŒS ©r   r�   ©rŠ   rƒ   s     rg   rŒ   z#per_category_ap.<locals>.<listcomp>;  s   € Ð!Ð!Ð!�aˆQˆqŒTÐ!Ð!Ð!r…   c                ó2   — g | ]}|d          �|d          nd‘ŒS )r   Nr   r�   r&  s     rg   rŒ   z#per_category_ap.<locals>.<listcomp><  s)   € Ð=Ð=Ð=°!�a˜”dÐ&ˆa�Œdˆd¨AÐ=Ð=Ð=r…   r�   r=   ç333333Ó?úaxes.prop_cyclerÏ   z
grid.colorz#ccccccc                ó$   •— g | ]}|d k    r‰n‰‘ŒS ©r#  r�   )rŠ   r®   Ú	bar_colorÚ	sep_colors     €€rg   rŒ   z#per_category_ap.<locals>.<listcomp>D  s%   ø€ ÐHÐHÐH¸Q˜q Ešz˜z�)�)¨yÐHÐHÐHr…   r  ©r  rÏ   c                ó,   — g | ]\  }}|d k    r|d›nd‘ŒS )r#  r.   Ú r�   ©rŠ   r®   Úvs      rg   rŒ   z#per_category_ap.<locals>.<listcomp>G  s,   € ÐVÐVÐV¹4¸1¸a A¨¢J J˜�j�j�j°BÐVÐVÐVr…   g      @r  r  c                ó$   — g | ]\  }}|d k    ¯|‘ŒS r+  r�   r1  s      rg   rŒ   z#per_category_ap.<locals>.<listcomp>K  s!   € ÐFÐFÐF™T˜Q ¸1Àº:¸:�q¸:¸:¸:r…   r   ú--r   z	Mean AP: ú.3f)Ú	linestyler0   r1   ÚAPr>   zlower rightrD   zPer-Category APrƒ   rÔ   )r   ru   rp   rÚ   ÚitemsrO   rR   r   r   rà   ÚnextÚiterÚrcParamsr  rN   Úzipr  r
   rÝ   Úaxvlinerã   rå   r  ræ   rX   r	   r   )r  r{   r  r   r   r   r   rZ   r[   r!  r8  ÚnamesÚvaluesr  r`   Úcolorsr  r  Úreal_valuesÚmean_apr,  r-  s                       @@rg   Úper_category_aprC    s  øø€ õB ‘”�I€CˆˆAà× Ò  ¨bÑ1Ô1€IØð MÝÐKÑLÔLÐLå�9—?’?Ñ$Ô$¨.¨.À$ÐGÑGÔG€Eå
ˆ5�z„z�E˜HÑ$Ò$Ð$Ø�f�u�f”  Ñ/°%¸¸	¸
¸
Ô2CÑCˆà!Ð!˜5Ð!Ñ!Ô!€EØ=Ð=°uÐ=Ñ=Ô=€FÝ�5‰zŒz€Hà	�Š�	 %¨Ñ4Ô4Ñ	5Ô	5ð ?ð ?Ý˜q¥# a¨¨x©Ñ"8Ô"8Ð9¸2Ñ>Ô>‰ˆˆRå�˜cœlÐ+<Ô=Ñ>Ô>Ñ?Ô?ÀÔHˆ	Ø”L×$Ò$ \°9Ñ=Ô=ˆ	ØHÐHÐHÐHÐHÀ%ÐHÑHÔHˆØ�wŠw•u˜X‘”¨°sÀ&ˆwÑIÔIˆàVÐVÅ3ÀuÈfÑCUÔCUÐVÑVÔVˆ
Ø
�Š�T *°sÀAˆÑFÔFÐFÝ   VÑ,Ô,Ð,àFÐF¥S¨°Ñ%7Ô%7ÐFÑFÔFˆØ9DÐK•#�kÑ"Ô"¥S¨Ñ%5Ô%5Ñ5Ð5È!ˆØ
�
Š
�7 d°aÐ?XÈ7Ð?XÐ?XÐ?Xˆ
ÑYÔYÐYà
�Š•e˜H‘o”oÑ&Ô&Ð&Ø
×Ò˜5Ñ!Ô!Ð!Ø
�ŠÑÔÐØ
�Š�dÑÔÐØ
�	Š	˜1 -ˆ	Ñ0Ô0Ð0Ý˜Ð-¸#Ð>Ñ>Ô>Ð>ð+?ð ?ð ?ñ ?ô ?ð ?ð ?ð ?ð ?ð ?ð ?øøøð ?ð ?ð ?ð ?õ. ˜C  YÑ/Ô/Ð/s   ÃG
J1Ê1J5Ê8J5)r   r   r   r   Ú	tide_dictc               ó  ‡— t          ¦   «         \  }}}| d         Š| d         }t          t          ¦  «        }ˆfd„|D ¦   «         }	|                     t	          ||¦  «        ¦  «        5  t          d|¦  «        \  }
}|                     t          t          |¦  «        ¦  «        |	d¬¦  «        }| 	                    |d„ |	D ¦   «         dd	¬
¦  «         t          ||	¦  «         |                     t          t          |¦  «        ¦  «        ¦  «         |                     |¦  «         |                     ¦   «          |                     d¦  «         t          |dd|d›�d¬¦  «         ddd¦  «         n# 1 swxY w Y   t!          |
||¦  «        S )aö  Plot TIDE error breakdown as horizontal bars.

    Parameters
    ----------
    tide_dict : dict
        Output of ``coco_eval.tide_errors()``.
    theme : str
        ``"cyanotype"`` (default) or ``"cyanotype-dark"``.
    paper_mode : bool
        White figure and axes background for PDF/LaTeX inclusion.
    ax : matplotlib.axes.Axes, optional
        Existing axes to draw on.
    save_path : str or Path, optional
        Save figure to this path.

    Returns
    -------
    (Figure, Axes)
    Údelta_apÚap_basec                ó<   •— g | ]}‰                      |d ¦  «        ‘ŒS )r~   ©ru   )rŠ   ÚerF  s     €rg   rŒ   ztide_errors.<locals>.<listcomp>z  s'   ø€ Ð8Ð8Ð8 qˆh�lŠl˜1˜cÑ"Ô"Ð8Ð8Ð8r…   )r�   r=   g333333ã?r  c                ó   — g | ]}|d ›‘ŒS )r5  r�   )rŠ   r2  s     rg   rŒ   ztide_errors.<locals>.<listcomp>  s   € Ð">Ð">Ð">°! a : :Ð">Ð">Ð">r…   r>   r  r  u   Î”APzTIDE Error Breakdownzbaseline AP=r5  rƒ   rG   N)r   rM   r   rR   r   r   r  rN   rO   r  r
   rã   rå   r  ræ   r	   r   )rD  r   r   r   r   rZ   r[   rG  Úerror_typesr?  r`   r  rF  s               @rg   Útide_errorsrM  Y  sÖ  ø€ õ6 ‘”�I€CˆˆAà˜Ô$€HØ˜	Ô"€GåÕ'Ñ(Ô(€KØ8Ð8Ð8Ð8¨KÐ8Ñ8Ô8€Fà	�Š�	 %¨Ñ4Ô4Ñ	5Ô	5ð 
kð 
kÝ˜f bÑ)Ô)‰ˆˆRØ�wŠw•u�S Ñ-Ô-Ñ.Ô.°¸sˆwÑCÔCˆØ
�Š�TÐ">Ð">°vÐ">Ñ">Ô">ÈÐTUˆÑVÔVÐVÝ   VÑ,Ô,Ð,à
�Š•e�C Ñ,Ô,Ñ-Ô-Ñ.Ô.Ð.Ø
×Ò˜;Ñ'Ô'Ð'Ø
�ŠÑÔÐØ
�Š�jÑ!Ô!Ð!Ý˜Ð2Ð=YÈGÐ=YÐ=YÐ=YÐfiÐjÑjÔjÐjð
kð 
kð 
kñ 
kô 
kð 
kð 
kð 
kð 
kð 
kð 
køøøð 
kð 
kð 
kð 
kõ ˜C  YÑ/Ô/Ð/s   Á)C=E2Å2E6Å9E6)Ún_binsÚiou_thresholdr   r   r   r   rN  rO  c               óœ  — ddl }t          ¦   «         \  }}	}	t          | t          ¦  «        r| }
n|                      ||¬¦  «        }
|
d         }|
d         }|
d         }|                     d„ |D ¦   «         ¦  «        }|                     d„ |D ¦   «         ¦  «        }|                     d	„ |D ¦   «         ¦  «        }|                     d
„ |D ¦   «         ¦  «        }|r|d         d         |d         d         z
  nd}|dk    }t          |¦  «        d         }|                     t          ||¦  «        ¦  «        5  |j	        }|d         |d         |d         }}}t          d|d¬¦  «        \  }}|                     ddgddgddgd|d¬¦  «         |                     ddgddgd|ddd¬¦  «         |                     ||         ||         |dz  ddd¬ ¦  «         ||         }||         }||         }||z
  }|dz  }d!}|dk    } |                      ¦   «         r0|                     ||          ||          ||          ||d"d#¬$¦  «         d%}|dk     }!|!                     ¦   «         r6|                     ||!         ||!          ||!         |d&|d'd|rdnd"d#¬(¦
  «
         |                     d)d)d*d+d¬,¦  «         |                     d-d.¬/¦  «         |                     d0d1d2|d3›d4|d3›�|j        d-d5d6|d7|d8         d|d9œ¬:¦	  «	         t'          |d;d<|
d=         d>›d?|
d@         dA›dB�dC¬D¦  «         ddd¦  «         n# 1 swxY w Y   t)          |||¦  «        S )Ea8  Plot a reliability diagram showing predicted confidence vs actual accuracy.

    Parameters
    ----------
    cal_or_eval : dict or COCOeval
        Either the output of ``ev.calibration()`` (a dict) or a ``COCOeval``
        instance (must have ``evaluate()`` called). If a ``COCOeval`` is
        passed, ``calibration()`` is called automatically.
    n_bins : int
        Number of bins (only used when ``cal_or_eval`` is a COCOeval).
    iou_threshold : float
        IoU threshold (only used when ``cal_or_eval`` is a COCOeval).
    theme : str
        ``"cyanotype"`` (default) or ``"cyanotype-dark"``.
    paper_mode : bool
        White figure and axes background for PDF/LaTeX inclusion.
    ax : matplotlib.axes.Axes, optional
    save_path : str or Path, optional

    Returns
    -------
    (Figure, Axes)
    r   N)rN  rO  ÚbinsÚeceÚmcec                ó6   — g | ]}|d          |d         z   dz  ‘ŒS )Ú	bin_lowerÚ	bin_upperr,   r�   ©rŠ   rÅ   s     rg   rŒ   z'reliability_diagram.<locals>.<listcomp>¹  s*   € ÐNÐNÐNÀA˜1˜[œ>¨A¨k¬NÑ:¸aÑ?ÐNÐNÐNr…   c                ó   — g | ]
}|d          ‘ŒS )Úavg_accuracyr�   rW  s     rg   rŒ   z'reliability_diagram.<locals>.<listcomp>º  s   € Ð;Ð;Ð;°˜1˜^Ô,Ð;Ð;Ð;r…   c                ó   — g | ]
}|d          ‘ŒS )Úavg_confidencer�   rW  s     rg   rŒ   z'reliability_diagram.<locals>.<listcomp>»  s   € Ð>Ð>Ð>°A˜AÐ.Ô/Ð>Ð>Ð>r…   c                ó   — g | ]
}|d          ‘ŒS )r  r�   rW  s     rg   rŒ   z'reliability_diagram.<locals>.<listcomp>¼  s   € Ð0Ð0Ð0 a�q˜”zÐ0Ð0Ð0r…   rV  rU  gš™™™™™¹?Úfpzxtick.colorúaxes.edgecolorz
text.colorr'   r)   r*   r   g¸…ëQ¸®?Ú
_nolegend_)ÚalpharÏ   r1   r4  ÚPerfect)rÏ   r0   r1   Úzorderg333333ë?r  ÚAccuracyr,   )Úwidthr`  r1   rb  FÚGapr  )Úbottomrd  rÏ   r1   rb  Tr¶   z///)rf  rd  Ú	facecolorÚ	edgecolorÚhatchr0   r1   rb  r3   r4   Ú
Confidencer7   r>   z
upper leftrD   gffffffî?gš™™™™™©?zECE = z.4fz
MCE = rÑ   rf  zround,pad=0.3zaxes.facecolor)Úboxstylerg  r`  rh  )r  r@   rÍ   rÎ   rÏ   ÚbboxzReliability Diagramu   IoUâ‰¥rO  r.   z, Únum_detectionsú,z detectionsrF   rG   )r‘   r   Ú
isinstanceÚdictÚcalibrationÚarrayr   rR   r   r;  r   Úfill_betweenrU   ÚbarÚanyrW   rX   rÀ   r  r	   r   )"Úcal_or_evalrN  rO  r   r   r   r   r•   rZ   r[   ÚcalrQ  rR  rS  Ú	midpointsÚ
accuraciesÚconfidencesr  Ú	bin_widthÚnonemptyÚ	gap_colorrþ   ÚtickÚspineÚinkr`   ÚmidÚaccÚconfÚgapsÚbar_wÚhas_gap_labelÚoverconfidentÚunderconfidents"                                     rg   Úreliability_diagramr‰  ‹  s„  € ðB ÐÐÐå‘”�I€CˆˆAå�+�tÑ$Ô$ð RØˆˆà×%Ò%¨VÀ=Ð%ÑQÔQˆàˆvŒ;€DØ
ˆeŒ*€CØ
ˆeŒ*€Cà—’ÐNÐNÈÐNÑNÔNÑOÔO€IØ—’Ð;Ð;°dÐ;Ñ;Ô;Ñ<Ô<€JØ—(’(Ð>Ð>¸Ð>Ñ>Ô>Ñ?Ô?€KØ�XŠXÐ0Ð0¨4Ð0Ñ0Ô0Ñ1Ô1€FØ?CÐL��Q”˜Ô$ t¨A¤w¨{Ô';Ñ;Ð;È€IØ˜Šz€Hõ ˜EÑ"Ô" 4Ô(€Ià	�Š�	 %¨Ñ4Ô4Ñ	5Ô	5ð \
ð \
ð
 Œ\ˆØ˜mÔ,¨bÐ1AÔ.BÀBÀ|ÔDT�Sˆeˆå˜f b°Ð>Ñ>Ô>‰ˆˆRð 	�Š˜˜A˜  A ¨¨A¨°dÀ$ÈlˆÑ[Ô[Ð[ð 	�Š��A�˜˜A˜ ¨D¸AÀYÐWXˆÑYÔYÐYð 	�Šˆy˜Ô" J¨xÔ$8À	ÈDÑ@PÐX[ÐcmÐvwˆÑxÔxÐxð ˜Ô!ˆØ˜Ô"ˆØ˜8Ô$ˆØ�c‰zˆØ˜DÑ ˆØˆà˜qšˆØ×ÒÑÔð 
	!Ø�FŠFØ�MÔ"Ø�]Ô#Ø˜=Ô)ØØØØð ñ ô ð ð !ˆMà šˆØ×ÒÑÔð 	Ø�FŠFØ�NÔ#Ø�nÔ%Ð%Ø˜NÔ+ØØ Ø#ØØØ&3Ð>�l�l¸Øð ñ ô ð ð 	�Š�F °ÀÐU_ˆÑ`Ô`Ð`Ø
�	Š	˜1 ,ˆ	Ñ/Ô/Ð/ð 	�ŠØØØ/�SÐ/Ð/Ð/ cÐ/Ð/Ð/Ø”lØØØØð ,ØÐ 0Ô1ØØ"ðð ð 	ñ 	
ô 	
ð 	
õ& 	ØØ!Øa  _Ô!5ÐaÐaÐa¸SÐAQÔ=RÐaÐaÐaÐaØð		
ñ 	
ô 	
ð 	
ðo\
ð \
ð \
ñ \
ô \
ð \
ð \
ð \
ð \
ð \
ð \
øøøð \
ð \
ð \
ð \
õ| ˜C  YÑ/Ô/Ð/s   Ä9G-L2Ì2L6Ì9L6Úcompare_resultc               ó&  ‡‡‡— ddl }t          ¦   «         \  }}}| d         Š| d         Š|                      d¦  «        Š|                      dd¦  «        }|                      dd	¦  «        }	|                      d
¦  «        }
|
rˆˆfd„|
D ¦   «         }
nt          ˆˆfd„‰D ¦   «         ¦  «        }
|                     ˆfd„|
D ¦   «         ¦  «        }|                     ˆfd„|
D ¦   «         ¦  «        }|                     t          |
¦  «        ¦  «        }d}|                     t          ||¦  «        ¦  «        5  t          t          ddt          |
¦  «        z  ¦  «        df|¦  «        \  }}|j        d                              ¦   «         d         }|d         }t          |¦  «        dk    r|d         n|d         }d}‰r||                     ˆfd„|
D ¦   «         ¦  «        }|                     ˆfd„|
D ¦   «         ¦  «        }|                     |||z   z
  ||z   |z
  g¦  «        }|                     |dd¦  «        }|                     ||dz  z
  ||||¬¦  «         |                     ||dz  z   |||	||dddi¬¦  «         |                     |¦  «         |                     |
ddd¬ ¦  «         |                     d!¦  «         |                     d¬"¦  «         t'          |d#d$¬%¦  «         ddd¦  «         n# 1 swxY w Y   t)          |||¦  «        S )&aÉ  Grouped bar chart comparing metrics between two models.

    Parameters
    ----------
    compare_result : dict
        Output of ``hotcoco.compare(eval_a, eval_b)``.
    theme : str
        ``"cyanotype"`` (default) or ``"cyanotype-dark"``.
    paper_mode : bool
        White figure and axes background for PDF/LaTeX inclusion.
    ax : matplotlib.axes.Axes, optional
    save_path : str or Path, optional

    Returns
    -------
    (Figure, Axes)
    r   NÚ	metrics_aÚ	metrics_bÚciÚname_aúModel AÚname_búModel BÚmetric_keysc                ó|   •— g | ]8}‰                      |d ¦  «        dk    s‰                      |d ¦  «        dk    ¯6|‘Œ9S )éÿÿÿÿr   rI  ©rŠ   rŽ   rŒ  r�  s     €€rg   rŒ   z"comparison_bar.<locals>.<listcomp>J  sK   ø€ ÐZÐZÐZ�a 9§=¢=°°BÑ#7Ô#7¸1Ò#<Ð#<À	ÇÂÈaÐQSÑ@TÔ@TÐXYÒ@YÐ@Y�Ð@YÐ@YÐ@Yr…   c              3  óh   •K  — | ],}‰|         d k    s‰                      |d¦  «        d k    ¯(|V — Œ-dS )r   r•  NrI  r–  s     €€rg   ú	<genexpr>z!comparison_bar.<locals>.<genexpr>L  sL   øè è € Ð]Ð]˜A¨i¸¬l¸aÒ.?Ð.?À9Ç=Â=ÐQRÐTVÑCWÔCWÐ[\ÒC\ÐC\�aÐC\ÐC\ÐC\ÐC\Ð]Ð]r…   c                óX   •— g | ]&}t          ‰                     |d ¦  «        d ¦  «        ‘Œ'S r%  ©rà   ru   )rŠ   rŽ   rŒ  s     €rg   rŒ   z"comparison_bar.<locals>.<listcomp>N  ó1   ø€ ÐAÐAÐA°q•s˜9Ÿ=š=¨¨AÑ.Ô.°Ñ2Ô2ÐAÐAÐAr…   c                óX   •— g | ]&}t          ‰                     |d ¦  «        d ¦  «        ‘Œ'S r%  rš  )rŠ   rŽ   r�  s     €rg   rŒ   z"comparison_bar.<locals>.<listcomp>O  r›  r…   r´   r�   çš™™™™™é?rË   r)  rÏ   r   c                ó8   •— g | ]}|‰v r‰|         d          nd‘ŒS )Úlowerr   r�   ©rŠ   rŽ   rŽ  s     €rg   rŒ   z"comparison_bar.<locals>.<listcomp>]  ó-   ø€ ÐJÐJÐJÀ¨Q°"¨W¨W˜2˜aœ5 œ>˜>¸!ÐJÐJÐJr…   c                ó8   •— g | ]}|‰v r‰|         d          nd‘ŒS )Úupperr   r�   r   s     €rg   rŒ   z"comparison_bar.<locals>.<listcomp>^  r¡  r…   r,   )r1   rÏ   r  r0   )r1   rÏ   ÚyerrÚcapsizeÚerror_kwrÐ   rÑ   r>   )rÒ   rÍ   r@   ÚScore)r@   zModel ComparisonrF   rÔ   )r‘   r   ru   rÚ   rr  ÚarangerO   rR   r   r   rà   r;  Úby_keyÚcliprt  râ   rä   rç   rX   r	   r   )rŠ  r   r   r   r   r•   rZ   r[   r�  r‘  rÛ   Úvals_aÚvals_brƒ   Ú	bar_widthr`   r@  Úcolor_aÚcolor_bÚyerr_bÚloÚhirŽ  rŒ  r�  s                         @@@rg   Úcomparison_barr³  %  s®  øøø€ ð2 ÐÐÐå‘”�I€CˆˆAà˜{Ô+€IØ˜{Ô+€IØ	×	Ò	˜DÑ	!Ô	!€BØ×Ò ¨)Ñ4Ô4€FØ×Ò ¨)Ñ4Ô4€Fà×Ò˜mÑ,Ô,€DØð ^ØZÐZÐZÐZÐZ˜4ÐZÑZÔZˆˆåÐ]Ð]Ð]Ð]Ð] Ð]Ñ]Ô]Ñ]Ô]ˆà�XŠXÐAÐAÐAÐA¸DÐAÑAÔAÑBÔB€FØ�XŠXÐAÐAÐAÐA¸DÐAÑAÔAÑBÔB€Fà
�	Š	•#�d‘)”)ÑÔ€AØ€Ià	�Š�	 %¨Ñ4Ô4Ñ	5Ô	5ð @ð @Ý�s 1 c­C°©I¬I¡oÑ6Ô6¸Ð:¸BÑ?Ô?‰ˆˆRØ”Ð/Ô0×7Ò7Ñ9Ô9¸'ÔBˆØ˜”)ˆÝ" 6™{œ{¨Qš˜�&˜”)�)°F¸1´Iˆð ˆØð 	.Ø—’ÐJÐJÐJÐJÀTÐJÑJÔJÑKÔKˆBØ—’ÐJÐJÐJÐJÀTÐJÑJÔJÑKÔKˆBà—X’X˜v¨°"©Ñ5°vÀ±{ÀVÑ7KÐLÑMÔMˆFØ—W’W˜V Q¨Ñ-Ô-ˆFà
�Šˆq�9˜q‘=Ñ  &¨)¸6ÈˆÑQÔQÐQØ
�ŠØ�	˜A‘ÑØØØØØØØ! 1Ð%ð 	ñ 		
ô 		
ð 		
ð 	�Š�aÑÔÐØ
×Ò˜4¨"°À1ÐÑEÔEÐEØ
�Š�gÑÔÐØ
�	Š	˜1ˆ	ÑÔÐÝ˜Ð.¸3Ð?Ñ?Ô?Ð?ð?@ð @ð @ñ @ô @ð @ð @ð @ð @ð @ð @øøøð @ð @ð @ð @õB ˜C  YÑ/Ô/Ð/s   Ä7F4K7Ë7K;Ë>K;)Útop_kr   r   r   r   r´  c          	     óv  ‡‡— t          ¦   «         \  }}}|                      dg ¦  «        }|st          d¦  «        ‚|                      dd¦  «        }	|                      dd¦  «        }
t          |¦  «        d|z  k    r|d|…         || d…         z   }nt	          |¦  «        }d	„ |D ¦   «         }d
„ |D ¦   «         }t          |¦  «        d         }|d         |d         cŠŠˆˆfd„|D ¦   «         }t          |¦  «        }|                     t          ||¦  «        ¦  «        5  t          dt          dd|z  ¦  «        f|¦  «        \  }}| 
                    t          |¦  «        |d|¬¦  «         |                     d|j        d         dd¬¦  «         |                     t          |¦  «        ¦  «         |                     |¦  «         |                     ¦   «          |                     d|
› d|	› d�¦  «         t%          |d|
› d|	› �d¬¦  «         ddd¦  «         n# 1 swxY w Y   t'          |||¦  «        S )aQ  Horizontal bar chart of per-category AP deltas between two models.

    Parameters
    ----------
    compare_result : dict
        Output of ``hotcoco.compare(eval_a, eval_b)``.
    top_k : int
        Number of categories to show from each end (top improvements + top
        regressions). Default 20.
    theme : str
        ``"cyanotype"`` (default) or ``"cyanotype-dark"``.
    paper_mode : bool
        White figure and axes background for PDF/LaTeX inclusion.
    ax : matplotlib.axes.Axes, optional
    save_path : str or Path, optional

    Returns
    -------
    (Figure, Axes)
    Úper_categoryz'No per-category data in compare result.r�  r�  r‘  r’  r,   Nc                ó   — g | ]
}|d          ‘ŒS )rw   r�   r½   s     rg   rŒ   z#category_deltas.<locals>.<listcomp>¥  s   € Ð*Ð*Ð*˜qˆQˆzŒ]Ð*Ð*Ð*r…   c                ó   — g | ]
}|d          ‘ŒS )Údeltar�   r½   s     rg   rŒ   z#category_deltas.<locals>.<listcomp>¦  s   € Ð(Ð(Ð(˜Qˆa�ŒjÐ(Ð(Ð(r…   Úseriesr   r�   c                ó$   •— g | ]}|d k     r‰n‰‘ŒS r%  r�   )rŠ   ÚdÚdownÚups     €€rg   rŒ   z#category_deltas.<locals>.<listcomp>¬  s%   ø€ Ð4Ð4Ð4¨�a˜!’e�eˆdˆd Ð4Ð4Ð4r…   r=   r(  r  r.  r   r^  r�  ú-)rÏ   r0   r6  z
AP Delta (u    âˆ’ ú)zPer-Category AP Deltaz vs rƒ   rG   )r   ru   rp   rO   rM   r   rR   r   r   rà   r  rN   r=  r;  rã   rå   r  ræ   r	   r   )rŠ  r´  r   r   r   r   rZ   r[   Úcatsr�  r‘  Úshownr>  Údeltasrº  r@  r  r`   r½  r¾  s                     @@rg   Úcategory_deltasrÄ  x  s¨  øø€ õ: ‘”�I€CˆˆAà×Ò˜n¨bÑ1Ô1€DØð DÝÐBÑCÔCÐCà×Ò ¨)Ñ4Ô4€FØ×Ò ¨)Ñ4Ô4€Fõ ˆ4�y„y�1�u‘9ÒÐØ�V�e�V”˜t U F G Gœ}Ñ,ˆˆå�T‘
”
ˆà*Ð* EÐ*Ñ*Ô*€EØ(Ð( %Ð(Ñ(Ô(€Fõ ˜ÑÔ˜xÔ(€FØ�aŒy˜& œ)€H€Dˆ"Ø4Ð4Ð4Ð4Ð4¨VÐ4Ñ4Ô4€FÝ�5‰zŒz€Hà	�Š�	 %¨Ñ4Ô4Ñ	5Ô	5ð 
gð 
gÝ˜q¥# a¨¨x©Ñ"8Ô"8Ð9¸2Ñ>Ô>‰ˆˆRà
�Š•�h‘” °¸6ˆÑBÔBÐBØ
�
Š
�1˜CœLÐ)9Ô:ÀcÐUXˆ
ÑYÔYÐYà
�Š•e˜H‘o”oÑ&Ô&Ð&Ø
×Ò˜5Ñ!Ô!Ð!Ø
�ŠÑÔÐØ
�ŠÐ< 6Ð<Ð<°6Ð<Ð<Ð<Ñ=Ô=Ð=Ý˜Ð3ÀÐ>UÐ>UÈVÐ>UÐ>UÐbeÐfÑfÔfÐfð
gð 
gð 
gñ 
gô 
gð 
gð 
gð 
gð 
gð 
gð 
gøøøð 
gð 
gð 
gð 
gõ ˜C  YÑ/Ô/Ð/s   Ä#C0HÈH#È&H#)r   r   r   r    r   r!   r   r    r   r"   r   r#   r$   r%   )rk   rl   rj   rm   r   r    r   r!   r   r    r   r"   r   r#   r$   r%   )ro   r|   r{   rl   rj   rm   r   r    r   r!   r   r    r   r"   r   r#   r$   r%   )r¢   r£   rŸ   r"   r{   r!   r    r¤   r¡   r¥   r   r    r   r"   r   r#   r$   r%   )r¢   r£   r{   rl   r   r    r   r"   r   r#   r$   r%   )r  r£   r{   rl   r  rl   r   r    r   r"   r   r#   r$   r%   )
rD  r£   r   r    r   r"   r   r#   r$   r%   )rN  rl   rO  rm   r   r    r   r"   r   r#   r$   r%   )
rŠ  r£   r   r    r   r"   r   r#   r$   r%   )rŠ  r£   r´  rl   r   r    r   r"   r   r#   r$   r%   )#Ú__doc__Ú
__future__r   Úpathlibr   Útypingr   Úcorer   r   r	   r
   r   r   r   r   r   r\   r   r   r   r   r   r   rh   rx   r›   rž   r  r  rC  rM  r‰  r³  rÄ  r�   r…   rg   ú<module>rÊ     sf  ððð ð #Ð "Ð "Ð "Ð "Ð "à Ð Ð Ð Ð Ð Ø Ð Ð Ð Ð Ð ð
ð 
ð 
ð 
ð 
ð 
ð 
ð 
ð 
ð 
ð 
ð 
ð 
ð 
ð 
ð 
ð 
ð 
ð 
ð 
ð 
ð 
ð Ð Ð Ð Ð Ð Ø =Ð =Ð =Ð =Ð =Ð =Ð =Ð =Ð =Ð =Ð =Ð =ð $(ØØØØØØ#'ðC0ð C0ð C0ð C0ð C0ð C0ðT ØØØØØØ#'ð:0ð :0ð :0ð :0ð :0ð :0ð@ !%ØØØØØØØØ#'ðJ0ð J0ð J0ð J0ð J0ð J0ð` ØØØØ
ØØØ
ØØØð<ð <ð <ð <ð <ðD ØØØ.2ØØØØ#'ðT0ð T0ð T0ð T0ð T0ð T0ðt ØØØØ#'ðM0ð M0ð M0ð M0ð M0ð M0ðf ØØØØØ#'ðG0ð G0ð G0ð G0ð G0ð G0ðZ ØØØ#'ð/0ð /0ð /0ð /0ð /0ð /0ðj ØØØØØ#'ðW0ð W0ð W0ð W0ð W0ð W0ðz ØØØ#'ðP0ð P0ð P0ð P0ð P0ð P0ðl ØØØØ#'ðC0ð C0ð C0ð C0ð C0ð C0ð C0ð C0r…   