^jjddlmZddlmZddlmZddlmZmZm Z ddl Z ddl mZ ddlmZddlmZddlmZdd lmZmZmZdd lmZdd lmZmZdd lmZm Z dd l!m"Z"erddl#Z$eGddZ%GddeZ&y)) annotations)deepcopy) dataclass) TYPE_CHECKINGAnycastN)pyplot)ORIENTED_BOX_COORDINATES) Detections) box_iou_batchmask_iou_batchoriented_box_iou_batch)LEGACY_COLOR_PALETTE)Metric MetricTarget)ObjectSizeCategoryget_detection_size_category)ensure_pandas_installedceZdZUdZded<eddZeddZeddZded<ded <d ed <d ed <d ed<ded<ded<ded<ddZ ddZ ddZ y)MeanAverageRecallResultux The results of the Mean Average Recall metric calculation. Defaults to `0` if no detections or targets were provided. Attributes: metric_target: the type of data used for the metric - boxes, masks or oriented bounding boxes. mAR_at_1: the Mean Average Recall, when considering only the top highest confidence detection for each class. mAR_at_10: the Mean Average Recall, when considering top 10 highest confidence detections for each class. mAR_at_100: the Mean Average Recall, when considering top 100 highest confidence detections for each class. recall_per_class: the recall scores per class and IoU threshold. Shape: `(num_target_classes, num_iou_thresholds)` max_detections: the array with maximum number of detections considered. iou_thresholds: the IoU thresholds used in the calculations. matched_classes: the class IDs of all matched classes. Corresponds to the rows of `recall_per_class`. small_objects: the Mean Average Recall metric results for small objects (area < 32²). medium_objects: the Mean Average Recall metric results for medium objects (32² ≤ area < 96²). large_objects: the Mean Average Recall metric results for large objects (area ≥ 96²). r metric_targetc2t|jdS)Nrfloat recall_scoresselfs r/var/www/ramen.bs-engineer-server.com/venv/lib/python3.12/site-packages/supervision/metrics/mean_average_recall.pymAR_at_1z MeanAverageRecallResult.mAR_at_1?T''*++c2t|jdS)Nrrs r mAR_at_10z!MeanAverageRecallResult.mAR_at_10Cr r!c2t|jdS)Nrrs r mAR_at_100z"MeanAverageRecallResult.mAR_at_100Gr r!npt.NDArray[np.float64]rrecall_per_classnpt.NDArray[np.int32]max_detectionsnpt.NDArray[np.float32]iou_thresholdsmatched_classeszMeanAverageRecallResult | None small_objectsmedium_objects large_objectsc|jjd|jd|jdd|jdd|j dd|j d|jd}|jjd k(r|d z }t|j|jD]\}}|d |d |d z }d }|j3|t|jjd d |z}|d|z }|j3|t|jjd d |z}|d|z }|j 3|t|j jd d |z}|d|z }|S)a Format as a pretty string. Example: ```pycon >>> import numpy as np >>> import supervision as sv >>> from supervision.metrics import MeanAverageRecall >>> predictions = sv.Detections( ... xyxy=np.array([[0, 0, 10, 10]]), ... class_id=np.array([0]), ... confidence=np.array([0.9]) ... ) >>> targets = sv.Detections( ... xyxy=np.array([[0, 0, 10, 10]]), ... class_id=np.array([0]) ... ) >>> mar_metric = MeanAverageRecall() >>> mar_result = mar_metric.update(predictions, targets).compute() >>> print(mar_result) # doctest: +ELLIPSIS MeanAverageRecallResult: Metric target: MetricTarget.BOXES mAR @ 1: 1.0000 mAR @ 10: 1.0000 mAR @ 100: 1.0000 max detections: [ 1 10 100] IoU thresh: [0.5 0.55 ... 0.95] mAR per class: 0: [1. ... 1.] ... Medium objects: MeanAverageRecallResult: Metric target: MetricTarget.BOXES mAR @ 1: 0.0000 ... ``` z: Metric target: z mAR @ 1: z.4fz mAR @ 10: z mAR @ 100: z max detections: z IoU thresh: z mAR per class: rz No results z z:  z Small objects: z Medium objects: z Large objects: ) __class____name__rrr$r'r+r-r)sizezipr.r/strreplacer0r1)rout_strclass_idrecall_of_classindentindenteds r__str__zMeanAverageRecallResult.__str__UsP~~&&'(#1123==-.>>#./??3/0#2234#2234     % % * ' 'G),  $"7"7*  < %Ho H:R'8; ;G <     )D$6$6 7 ? ?6(m TTH +H:6 6G    *D$7$7 8 @ @F8} UUH ,XJ7 7G    )D$6$6 7 ? ?6(m TTH +H:6 6Gr!c4tddl}|j|j|jd}|j :|j j }|jD] \}}||d|<|j:|jj }|jD] \}}||d|<|j:|jj }|jD] \}}||d|<|j|dgS)zt Convert the result to a pandas DataFrame. Returns: The result as a DataFrame. rNzmAR @ 1zmAR @ 10z mAR @ 100small_objects_medium_objects_large_objects_)index) rpandasrr$r'r/ to_pandasitemsr0r1 DataFrame)rpd pandas_datasmall_objects_dfkeyvaluemedium_objects_dflarge_objects_dfs rrGz!MeanAverageRecallResult.to_pandass. !}}    )#11;;= .446 < U6; nSE23 <    * $ 3 3 = = ? /557 = U7< ocU34 =    )#11;;= .446 < U6; nSE23 <r||Ks33r!c gd}|j|j|jg}tdgdz}|jI|j}|gdz }||j|j|jgz }|tdgdzz }|j I|j }|gdz }||j|j|jgz }|tdgdzz }|j I|j }|gdz }||j|j|jgz }|td gdzz }d tjd <tjd \}}|jdd|jddd|jjd} |j| dtt!|} |j#| ||d} |j%| |j'|dd| D]O} | j)} |j+| j-| j/dz z| dz| dddQdtjd <tj0tj2y)z Plot the Mean Average Recall results. ![example_plot]( https://media.roboflow.com/supervision-docs/metrics/mAR_plot_example.png ){ align=center width="800" } rArN)zSmall: mAR @ 1zSmall: mAR @ 10zSmall: mAR @ 100)zMedium: mAR @ 1zMedium: mAR @ 10zMedium: mAR @ 100r&)zLarge: mAR @ 1zLarge: mAR @ 10zLarge: mAR @ 100 monospacez font.family) )figsizer#Valuebold) fontweightz-Mean Average Recall, by Object Size (target: )center)coloralign-right)rotationhag{Gz?z.2fbottom)rbvaz sans-serif)rr$r'rr/r0r1pltrcParamssubplotsset_ylim set_ylabelrrN set_titlerangelenbar set_xticksset_xticklabels get_heighttextget_x get_width tight_layoutshow)rlabelsvaluescolorsr/r0r1_axtitle x_positionsbarsrmy_values rplotzMeanAverageRecallResult.plots6--A&q)*Q.    ) ..M O OF &&''(( F +A./!3 3F    *!00N R RF ''(()) F +A./!3 3F    ) ..M O OF &&''(( F +A./!3 3F&1 ]# W-2 Aq g& 1Gr!rc(eZdZdZej f ddZddZ ddZddZ ddZ ddZ e ddZ e dd Ze dd Zdd Zdd Z dd Z ddZy)MeanAverageRecalla< Mean Average Recall (mAR) measures how well the model detects and retrieves relevant objects by averaging recall over multiple IoU thresholds, classes and detection limits. Intuitively, while Recall measures the ability to find all relevant objects, mAR narrows down how many detections are considered for each class. For example, mAR @ 100 considers the top 100 highest confidence detections for each class. mAR @ 1 considers only the highest confidence detection for each class. Examples: ```pycon >>> import numpy as np >>> import supervision as sv >>> from supervision.metrics import MeanAverageRecall >>> predictions = sv.Detections( ... xyxy=np.array([[0, 0, 10, 10]]), ... class_id=np.array([0]), ... confidence=np.array([0.9]) ... ) >>> targets = sv.Detections( ... xyxy=np.array([[0, 0, 10, 10]]), ... class_id=np.array([0]) ... ) >>> mar_metric = MeanAverageRecall() >>> mar_result = mar_metric.update(predictions, targets).compute() >>> round(float(mar_result.mAR_at_100), 2) 1.0 ``` ![example_plot]( https://media.roboflow.com/supervision-docs/metrics/mAR_plot_example.png ){ align=center width="800" } cf||_g|_g|_tjgd|_y)z Initialize the Mean Average Recall metric. Args: metric_target: The type of detection data to use. )r#rUdN)_metric_target_predictions_list _targets_listnparrayr+)rrs r__init__zMeanAverageRecall.__init__)s-,35/1 hh|4r!c g|_g|_y)zR Reset the metric to its initial state, clearing all stored data. N)rrrs rresetzMeanAverageRecall.reset:s"$r!c4t|ts|g}t|ts|g}t|t|k7r$tdt|dt|d|jj ||j j ||S)a Add new predictions and targets to the metric, but do not compute the result. Args: predictions: The predicted detections. targets: The target detections. Returns: The updated metric instance. zThe number of predictions (z) and targets (z%) during the update must be the same.) isinstancelistrl ValueErrorrextendr)r predictionstargetss rupdatezMeanAverageRecall.updateAs+t,&-K'4(iG { s7| +-c+.>-?@ \N*OQ  %%k2 !!'* r!c,|j|j|j}|j|j|jtj \}}|j|||_|j|j|jtj\}}|j|||_|j|j|jtj\}}|j|||_ |S)z Calculate the Mean Average Recall metric based on the stored predictions and ground-truth, at different IoU thresholds and maximum detection counts. Returns: The Mean Average Recall metric result. ) _computerr'_filter_predictions_and_targets_by_sizerSMALLr/MEDIUMr0LARGEr1)rresultsmall_predictions small_targetsmedium_predictionsmedium_targetslarge_predictions large_targetss rcomputezMeanAverageRecall.compute`st55t7I7IJ+/+W+W  " "D$6$68J8P8P, (= $}}-> N  8 8&&(:(: N r!c Ztjddd}g}t||D]\}}|j|}|j|}t |dkDs8t |dk(rw|j tj d|jfttj dttj dt|jf|jtjk(r t||} n_|jtjk(r t!||} n5|jtj"k(r t%||} n t'd|j)|j |jntj*g|j |jntj*g| |} tj,t/t0j2tj4|j6 } |j | | tj8t |t/t0j2tj:|j| t/t0j2tj:|jf|st=|jtj |j>dtj d|j>df|j@|tj*gtddd St|D cgc]} tjB| d} } |jD| \}}}t=|j|||j@||ddd Scc} w) Ng?gffffff?rUrdtype)rz-Unsupported metric target for IoU calculation) rrr)r+r-r.r/r0r1)#rlinspacer7_detections_contentrlappendzerosr6boolintr;rrBOXESr MASKSr ORIENTED_BOUNDING_BOXESrr_match_detection_batchrargsortrnptNDArrayfloat32 confidencearangeint32rshaper+ concatenate#_compute_average_recall_for_classes)rpredictions_list targets_listr-statsrrprediction_contentstarget_contentsioumatchessorted_indicesrHconcatenated_statsrecall_scores_per_kr)unique_classess rrzMeanAverageRecall._compute}s*S$3$'(8,$G4  K"&":":;"G "66w?O7|a{#q(LLHHa)<)<%=TJHHT5HHT5#,, **l.@.@@+O=PQ,, 0B0BB,_>QR,, 0T0TT4+-@)K#99&//;$,,XXb\"++7 ((XXb\& G&(ZZckk"**5{7M7MNN&NLL#N3IIc+&67 RXX!6 8L8LM .!RXX!68H8HI  W4 l*"11 hh~';';A'>?!#1n.B.B1.E*F!G#22- "3 7"#"  EHKP5bnnUA6PP 4D 4 46H I >-~'----..)*  QsN(cptj|d\}}g}|jD]D}|j|||k|||k||} |j | } |j | Ftj |}tj|d} tj| d} |  |fS)NT) return_countsr&axisr#)runiquer+_compute_confusion_matrix_compute_recallrrmean) rrprediction_indicesprediction_class_idstrue_class_idsr class_counts recalls_at_kr+confusion_matrixr)average_recall_per_classrs rrz5MeanAverageRecall._compute_average_recall_for_classess(*yyt'T$  "11 2N#==*^;<$%7.%HI    $334DE     0 1 2xx - #%77>r!c|jd|jd}}tj||ft}|dddf|k(}t |D] \}} tj || k\|z} | djds5tj | d} || dddf} tj| | g} | djddkDrh| | dddfjddd} | tj| dddfdd} | tj| dddfdd} d|| dddfjt|f<|}|S) Nrrr#rr&T) return_index) rrrr enumeratewherestackhstackrrastyper)predictions_classestarget_classesrr-num_predictionsnum_iou_levelscorrect correct_classi iou_levelmatched_indicescombined_indices iou_valuesrresult_corrects rrz(MeanAverageRecall._match_detection_batchs~ % %a (   #(((O^H#I.H&llnO!#!#.!9"$((>":"$''.("CQ!#.!9"$''./"J"$((>":"0":!0!4!4Q!7#$#6";";A">"*^";*,((/B+ Y ') 0OOL    ,"4"4 4*6@oo ##++- -   ,"F"F F//%%&>?C3s8a<68hhs"**6U !!++- -243F3F2GHIIr!c|jtjk(r'tjdtj }|S|jtj k(rtjdt}|S|jtjk(r'tjdtj }|Std|j)N)rrSr)rrr)rrSr&r) rrrremptyrrrrr)r empty_boxes empty_masks empty_obbs rrz%MeanAverageRecall._make_empty_contents   ,"4"4 43588F"**3UK    ,"4"4 413)41PK    ,"F"F F13)2::1VI 243F3F2GHIIr!c~t|}|js|tjk(r|St ||j }||j k(}|j||_|j|j||_|j|j||_ |j|j||_ |j|j||_ |jG|jjD]*\}}tj|||j|<,|S)zBReturn a copy of detections with contents filtered by object size.)ris_emptyrANYrrrNrrr;r tracker_idr rHrr)rr  size_categorynew_detectionssizes size_maskrMrNs r_filter_detections_by_sizez,MeanAverageRecall._filter_detections_by_sizes4"*-    M5G5K5K$K! !+ND:X 0X @PX X t"?&"?2"?4 "? . "?   "?H2-%0   <<'-<'%:<'.<', <' ! <'<'|1 6J$ J$5G 46*6'6* 6 3 6r!r)' __future__rcopyr dataclassesrtypingrrrnumpyr numpy.typingr matplotlibr resupervision.configr supervision.detection.corer 'supervision.detection.utils.iou_and_nmsr r rsupervision.draw.colorrsupervision.metrics.corerr%supervision.metrics.utils.object_sizerrsupervision.metrics.utils.utilsrrFrJrrrr!rr4sq"!++$71 89D aa aH66r!