§
    >²¸j:ã  ã                  óX  — d dl mZ d dlZd dlmZmZ d dlmZmZ d dl	m
Z
mZ d dlZd dlm	Z d dlmZ d dlmZ d dlmZmZ d d	lmZmZmZ d d
lmZmZ d dlmZ d dl m!Z!  ej"        e#¦  «        Z$e%e&z  e'e%         z  e'e(         z  ej)        ej*                 z  ej)        ej+                 z  Z,e-e,e,f         Z.e%ej/        e
         z  ej)        ej0                 z  e'e
         z  e&z  Z1ej)        ej0                 e'e
         z  e&z  Z2dd„Z3dd„Z4 ed¬¦  «         G d„ d¦  «        ¦   «         Z5dS )é    )ÚannotationsN)ÚIterableÚIterator)Ú	dataclassÚfield)ÚAnyÚcast)ÚCLASS_NAME_DATA_FIELD)Ú
Detections)Ú_DetectionDataTypeÚ_DetectionDataValueType)Úget_data_itemÚis_data_equalÚ
merge_data)ÚOverlapMetricÚbox_non_max_suppression)Úwarn_deprecated)Ú_validate_keypoints_fieldsÚfirstúnpt.NDArray[np.generic] | NoneÚsecondÚreturnÚboolc                óD   — | �|€| d u o|d u S t          j        | |¦  «        S ©N)ÚnpÚarray_equal)r   r   s     úk/root/private/project/pallet_detection_v2/.venv/lib/python3.11/site-packages/supervision/key_points/core.pyÚ_optional_array_equalr   '   s2   € ð €}˜˜Ø˜ˆ}Ð/ ¨4 Ð/ÝŒ>˜% Ñ(Ô(Ð(ó    ÚiÚ_RowIndexInputÚ_NormalizedRowIndexc                óä  — t          | t          t          j        f¦  «        r5t	          t
          t          j        t          | ¦  «        g¦  «        ¦  «        S t          | t          j        ¦  «        rI| j        t          k    r9t	          t
          t          j
        |                      ¦   «         ¦  «        ¦  «        S t          | t          j        ¦  «        r3| j        dk    r(t	          t
          |                      d¦  «        ¦  «        S t          | t          ¦  «        rT| rRt          d„ | D ¦   «         ¦  «        r9t	          t
          t          j
        t          j        | ¦  «        ¦  «        ¦  «        S | S )a�  Normalise *i* to a 1-D row index for 1-D per-object fields.

    Handles:
    - Python int or np.integer scalar  -> np.array([int(i)])
    - boolean np.ndarray (any shape)   -> np.flatnonzero(i.ravel())
    - non-bool 0-d np.ndarray          -> reshaped to shape (1,)
    - list of bool                     -> np.flatnonzero(np.array(i))
    - slice, list of ints, 1-D ndarray -> returned as-is
    r   é   c              3  ó@   K  — | ]}t          |t          ¦  «        V — Œd S r   ©Ú
isinstancer   ©Ú.0Úxs     r   ú	<genexpr>z'_normalize_row_index.<locals>.<genexpr>B   s,   è è € Ð(HÐ(HÀ­°AµtÑ)<Ô)<Ð(HÐ(HÐ(HÐ(HÐ(HÐ(Hr    )r(   Úintr   Úintegerr	   r#   ÚarrayÚndarrayÚdtyper   ÚflatnonzeroÚravelÚndimÚreshapeÚlistÚall)r!   s    r   Ú_normalize_row_indexr8   0   s  € õ �!•c�2œ:Ð&Ñ'Ô'ð =ÝÕ'­¬µ3°q±6´6°(Ñ);Ô);Ñ<Ô<Ð<Ý�!•R”ZÑ Ô ð D Q¤Wµ¢_ _ÝÕ'­¬¸¿º¹	¼	Ñ)BÔ)BÑCÔCÐCÝ�!•R”ZÑ Ô ð 7 Q¤V¨q¢[ [ÝÕ'¨¯ª°1©¬Ñ6Ô6Ð6Ý�!•TÑÔð F˜qð F¥SÐ(HÐ(HÀaÐ(HÑ(HÔ(HÑ%HÔ%Hð FÝÕ'­¬½¼À¹¼Ñ)DÔ)DÑEÔEÐEØ€Hr    F)Úinitc                  ó  — e Zd ZU dZded<   dZded<   dZded<   dZded	<   dZd
ed<    e	e
¬¦  «        Zded<   	 	 	 	 	 dPddœdQd„ZdRd„ZedSd„¦   «         Zej        dTd„¦   «         ZdUd„ZdVd„ZdWd „ZedXd#„¦   «         ZedYd'„¦   «         ZedZd)„¦   «         Zed[d+„¦   «         Zed\d-„¦   «         Zed]d/„¦   «         Zd^d2„Zd_d6„Zd`d9„Zdad<„Zdbd>„Zedcd?„¦   «         Z ddd@„Z!ededC„¦   «         Z"dDdEe#j$        fdfdK„Z%	 dgdhdO„Z&dS )iÚ	KeyPointsu‡  
    The `sv.KeyPoints` class in the Supervision library standardizes results from
    various keypoint detection and pose estimation models into a consistent format. This
    class simplifies data manipulation and filtering, providing a uniform API for
    integration with Supervision [keypoints annotators](/latest/keypoint/annotators).

    === "RF-DETR"

        [RF-DETR](https://github.com/roboflow/rf-detr) keypoint models return
        `sv.KeyPoints` directly from `model.predict()` â€” no additional
        conversion is needed.

        ```python
        from supervision import _cv2 as cv2
        import supervision as sv
        from rfdetr import RFDETRKeypointPreview

        image = cv2.imread("<SOURCE_IMAGE_PATH>")
        model = RFDETRKeypointPreview()

        key_points = model.predict(image)
        ```

    === "Ultralytics"

        Use [`sv.KeyPoints.from_ultralytics`](/latest/keypoint/core/#supervision.key_points.core.KeyPoints.from_ultralytics)
        method, which accepts [YOLOv8-pose](https://docs.ultralytics.com/models/yolov8/), [YOLO11-pose](https://docs.ultralytics.com/models/yolo11/)
        [pose](https://docs.ultralytics.com/tasks/pose/) result.

        ```python
        from supervision import _cv2 as cv2
        import supervision as sv
        from ultralytics import YOLO

        image = cv2.imread("<SOURCE_IMAGE_PATH>")
        model = YOLO('yolo11s-pose.pt')

        result = model(image)[0]
        key_points = sv.KeyPoints.from_ultralytics(result)
        ```

    === "Inference"

        Use [`sv.KeyPoints.from_inference`](/latest/keypoint/core/#supervision.key_points.core.KeyPoints.from_inference)
        method, which accepts [Inference](https://inference.roboflow.com/) pose result.

        ```python
        from supervision import _cv2 as cv2
        import supervision as sv
        from inference import get_model

        image = cv2.imread("<SOURCE_IMAGE_PATH>")
        model = get_model(model_id="<POSE_MODEL_ID>", api_key="<ROBOFLOW_API_KEY>")

        result = model.infer(image)[0]
        key_points = sv.KeyPoints.from_inference(result)
        ```

    === "MediaPipe"

        Use [`sv.KeyPoints.from_mediapipe`](/latest/keypoint/core/#supervision.key_points.core.KeyPoints.from_mediapipe)
        method, which accepts [MediaPipe](https://github.com/google-ai-edge/mediapipe)
        pose result.


        ```python
        from supervision import _cv2 as cv2
        import mediapipe as mp
        import supervision as sv

        image = cv2.imread("<SOURCE_IMAGE_PATH>")
        image_height, image_width, _ = image.shape
        mediapipe_image = mp.Image(
            image_format=mp.ImageFormat.SRGB,
            data=cv2.cvtColor(image, cv2.COLOR_BGR2RGB))

        options = mp.tasks.vision.PoseLandmarkerOptions(
            base_options=mp.tasks.BaseOptions(
                model_asset_path="pose_landmarker_heavy.task"
            ),
            running_mode=mp.tasks.vision.RunningMode.IMAGE,
            num_poses=2)

        PoseLandmarker = mp.tasks.vision.PoseLandmarker
        with PoseLandmarker.create_from_options(options) as landmarker:
            pose_landmarker_result = landmarker.detect(mediapipe_image)

        key_points = sv.KeyPoints.from_mediapipe(
            pose_landmarker_result, (image_width, image_height))
        ```

    === "Transformers"

        Use [`sv.KeyPoints.from_transformers`](/latest/keypoint/core/#supervision.key_points.core.KeyPoints.from_transformers)
        method, which accepts [ViTPose](https://huggingface.co/docs/transformers/en/model_doc/vitpose) result.

        ```python
        from PIL import Image
        import requests
        import supervision as sv
        import torch
        from transformers import (
            AutoProcessor,
            RTDetrForObjectDetection,
            VitPoseForPoseEstimation,
        )

        device = "cuda" if torch.cuda.is_available() else "cpu"
        image = Image.open("<SOURCE_IMAGE_PATH>")

        DETECTION_MODEL_ID = "PekingU/rtdetr_r50vd_coco_o365"

        detection_processor = AutoProcessor.from_pretrained(DETECTION_MODEL_ID, use_fast=True)
        detection_model = RTDetrForObjectDetection.from_pretrained(DETECTION_MODEL_ID, device_map=DEVICE)

        inputs = detection_processor(images=frame, return_tensors="pt").to(DEVICE)

        with torch.no_grad():
            outputs = detection_model(**inputs)

        target_size = torch.tensor([(frame.height, frame.width)])
        results = detection_processor.post_process_object_detection(
            outputs, target_sizes=target_size, threshold=0.3)

        detections = sv.Detections.from_transformers(results[0])
        boxes = sv.xyxy_to_xywh(detections[detections.class_id == 0].xyxy)

        POSE_ESTIMATION_MODEL_ID = "usyd-community/vitpose-base-simple"

        pose_estimation_processor = AutoProcessor.from_pretrained(POSE_ESTIMATION_MODEL_ID)
        pose_estimation_model = VitPoseForPoseEstimation.from_pretrained(
            POSE_ESTIMATION_MODEL_ID, device_map=DEVICE)

        inputs = pose_estimation_processor(frame, boxes=[boxes], return_tensors="pt").to(DEVICE)

        with torch.no_grad():
            outputs = pose_estimation_model(**inputs)

        results = pose_estimation_processor.post_process_pose_estimation(outputs, boxes=[boxes])
        key_point = sv.KeyPoints.from_transformers(results[0])
        ```

    Attributes:
        xy: An array of shape `(n, m, 2)` containing
            `n` detected objects, each composed of `m` equally-sized
            sets of key points, where each point is `[x, y]`.
        class_id: An array of shape
            `(n,)` containing the class ids of the detected objects.
        keypoint_confidence: An array of shape
            `(n, m)` containing the confidence scores of each keypoint.
        detection_confidence: An array of shape
            `(n,)` containing the detection-level confidence scores.
        visible: An optional boolean array of shape
            `(n, m)` indicating which keypoints are visible. When ``None``,
            all keypoints are treated as visible. Set this to filter anchors
            without removing data: ``key_points.visible = key_points.keypoint_confidence > 0.3``.
        data: A dictionary containing additional
            data where each key is a string representing the data type, and the value
            is either a NumPy array or a list of corresponding data of length `n`
            (one entry per detected object).
    únpt.NDArray[np.float32]ÚxyNúnpt.NDArray[np.int_] | NoneÚclass_idúnpt.NDArray[np.float32] | NoneÚkeypoint_confidenceÚdetection_confidenceúnpt.NDArray[np.bool_] | NoneÚvisible)Údefault_factoryr   Údata)Ú
confidenceú_DetectionDataType | NonerG   r   ÚNonec               óÒ   — |�"|�t          d¦  «        ‚t          d¦  «         |}|| _        || _        || _        || _        || _        |�|ni | _        |                      ¦   «          dS )a÷  Initialize KeyPoints.

        Args:
            xy: Array of shape `(n, m, 2)` with keypoint coordinates.
            class_id: Array of shape `(n,)` with class IDs. Defaults to None.
            keypoint_confidence: Array of shape `(n, m)` with per-keypoint
                confidence scores. Defaults to None.
            detection_confidence: Array of shape `(n,)` with detection-level
                confidence scores. Defaults to None.
            visible: Boolean array of shape `(n, m)` indicating visible
                keypoints. Defaults to None.
            data: Dictionary of additional per-detection data arrays.
                Defaults to an empty dict.
            confidence: Deprecated since `0.29.0`, removed in `0.32.0`.
                Use ``keypoint_confidence`` instead. Raises ``ValueError``
                if passed together with ``keypoint_confidence``.

        Raises:
            ValueError: If both ``confidence`` and ``keypoint_confidence``
                are provided.
        Nuw   Cannot pass both 'confidence' and 'keypoint_confidence'. 'confidence' is deprecated â€” use 'keypoint_confidence' only.zˆ'confidence' parameter in `KeyPoints()` is deprecated since `0.29.0` and will be removed in `0.32.0`. Use 'keypoint_confidence' instead.)	Ú
ValueErrorr   r=   r?   rA   rB   rD   rF   Ú__post_init__)Úselfr=   r?   rA   rB   rD   rF   rG   s           r   Ú__init__zKeyPoints.__init__ò   s—   € ð@ Ð!Ø"Ð.Ý ðUñô ð õ ð1ñô ð ð
 #-ÐàˆŒØ ˆŒØ#6ˆÔ Ø$8ˆÔ!ØˆŒØ Ð,�D�D°"ˆŒ	Ø×ÒÑÔÐÐÐr    c                ól   — t          | j        | j        | j        | j        | j        | j        ¬¦  «         d S )N©r=   r?   rG   rB   rD   rF   )r   r=   r?   rA   rB   rD   rF   ©rM   s    r   rL   zKeyPoints.__post_init__'  sC   € Ý"ØŒwØ”]ØÔ/Ø!%Ô!:Ø”LØ”ð	
ñ 	
ô 	
ð 	
ð 	
ð 	
r    c                ó.   — t          d¦  «         | j        S )z=Deprecated since 0.29.0. Use ``keypoint_confidence`` instead.ú}'KeyPoints.confidence' is deprecated since 0.29.0 and will be removed in 0.32.0. Use 'KeyPoints.keypoint_confidence' instead.©r   rA   rQ   s    r   rG   zKeyPoints.confidence1  s'   € õ 	ðNñ	
ô 	
ð 	
ð Ô'Ð'r    Úvaluec                ó2   — t          d¦  «         || _        d S )NrS   rT   )rM   rU   s     r   rG   zKeyPoints.confidence:  s*   € åðNñ	
ô 	
ð 	
ð $)ˆÔ Ð Ð r    r-   c                ó*   — t          | j        ¦  «        S )a­  
        Returns the number of objects in the `sv.KeyPoints` object.

        Returns:
            The number of objects.

        Example:
            ```pycon
            >>> import numpy as np
            >>> import supervision as sv
            >>> xy = np.array([[[10, 20], [30, 40]]], dtype=np.float32)
            >>> key_points = sv.KeyPoints(xy=xy)
            >>> len(key_points)
            1

            ```
        )Úlenr=   rQ   s    r   Ú__len__zKeyPoints.__len__B  s   € õ$ �4”7‰|Œ|Ðr    úyIterator[tuple[npt.NDArray[np.float32], npt.NDArray[np.float32] | None, npt.NDArray[np.int_] | None, _DetectionDataType]]c              #  óî   K  — t          t          | j        ¦  «        ¦  «        D ]P}| j        |         | j        �| j        |         nd| j        �| j        |         ndt          | j        |¦  «        fV — ŒQdS )z—
        Iterates over the Keypoint object and yield a tuple of
        `(xy, keypoint_confidence, class_id, data)` for each object detection.
        N)ÚrangerX   r=   rA   r?   r   rF   )rM   r!   s     r   Ú__iter__zKeyPoints.__iter__V  s”   è è € õ •s˜4œ7‘|”|Ñ$Ô$ð 	ð 	ˆAà”˜”
àÔ+Ð7ð Ô(¨Ô+Ð+àØ$(¤MÐ$=�”˜aÔ Ð À4Ý˜dœi¨Ñ+Ô+ðð ð ð ð ð	ð 	r    ÚotherÚobjectr   c                óŽ  — t          |t          ¦  «        st          S t          t	          j        | j        |j        ¦  «        t          | j        |j        ¦  «        t          | j	        |j	        ¦  «        t          | j
        |j
        ¦  «        t          | j        |j        ¦  «        t          | j        |j        ¦  «        g¦  «        S r   )r(   r;   ÚNotImplementedr7   r   r   r=   r   r?   rA   rB   rD   r   rF   )rM   r^   s     r   Ú__eq__zKeyPoints.__eq__n  s«   € Ý˜%¥Ñ+Ô+ð 	"Ý!Ð!Ýå”˜tœw¨¬Ñ1Ô1Ý% d¤m°U´^ÑDÔDÝ%ØÔ,¨eÔ.Gñô õ &ØÔ-¨uÔ/Iñô õ & d¤l°E´MÑBÔBÝ˜dœi¨¬Ñ4Ô4ðñ
ô 
ð 	
r    Úinference_resultr   c                óð  — t          |t          ¦  «        rt          d¦  «        ‚t          |d¦  «        r|                     dd¬¦  «        }n$t          |d¦  «        r|                     ¦   «         }|                     d¦  «        s|                      ¦   «         S |d         }d„ |D ¦   «         }t          d„ |D ¦   «         d	¬
¦  «        }t          j
        t          |¦  «        |dft          j        ¬¦  «        }t          j
        t          |¦  «        |ft          j        ¬¦  «        }t          t          ||¦  «        ¦  «        D ]F\  }\  }}	t          |	|d         ¦  «        D ]'\  }
}|d         |d         f|||
f<   |d         |||
f<   Œ(ŒGt          j        d„ |D ¦   «         ¦  «        }t          j        d„ |D ¦   «         ¦  «        }t           |i} | |||                     t$          ¦  «        |¬¦  «        S )aÂ  
        Create a `sv.KeyPoints` object from the [Roboflow](https://roboflow.com/)
        API inference result or the [Inference](https://inference.roboflow.com/)
        package results.

        Args:
            inference_result: The result from the
                Roboflow API or Inference package containing predictions with keypoints.

        Returns:
            A `sv.KeyPoints` object containing the keypoint coordinates, class IDs,
                class names, and confidences of each keypoint.

        Examples:
            ```python
            from supervision import _cv2 as cv2
            import supervision as sv
            from inference import get_model

            image = cv2.imread("<SOURCE_IMAGE_PATH>")
            model = get_model(model_id="<POSE_MODEL_ID>", api_key="<ROBOFLOW_API_KEY>")

            result = model.infer(image)[0]
            key_points = sv.KeyPoints.from_inference(result)
            ```

            ```python
            from supervision import _cv2 as cv2
            import supervision as sv
            from inference_sdk import InferenceHTTPClient

            image = cv2.imread("<SOURCE_IMAGE_PATH>")
            client = InferenceHTTPClient(
                api_url="https://detect.roboflow.com",
                api_key="<ROBOFLOW_API_KEY>"
            )

            result = client.infer(image, model_id="<POSE_MODEL_ID>")
            key_points = sv.KeyPoints.from_inference(result)
            ```
        z}from_inference() operates on a single result at a time.You can retrieve it like so:  inference_result = model.infer(image)[0]ÚdictT)Úexclude_noneÚby_aliasÚjsonÚpredictionsc                óL   — g | ]!}d „ t          |d         ¦  «        D ¦   «         ‘Œ"S )c                ó@   — g | ]\  }}|                      d |¦  «        ‘ŒS ©r?   )Úget)r*   ÚpositionÚkeypoints      r   ú
<listcomp>z7KeyPoints.from_inference.<locals>.<listcomp>.<listcomp>Á  s:   € ð ð ð á&�H˜hð —’˜Z¨Ñ2Ô2ðð ð r    Ú	keypoints)Ú	enumerate©r*   Ú
predictions     r   rp   z,KeyPoints.from_inference.<locals>.<listcomp>À  sN   € ð 
ð 
ð 
ð
 ð	ð å*3°J¸{Ô4KÑ*LÔ*Lðñ ô ð
ð 
ð 
r    c              3  ó>   K  — | ]}|¯t          |¦  «        d z   V — ŒdS )r%   N)Úmax)r*   Úobject_slotss     r   r,   z+KeyPoints.from_inference.<locals>.<genexpr>È  s4   è è € ÐMÐM |ÀÐM�S�ÑÔ Ñ"ÐMÐMÐMÐMÐMÐMr    r   )Údefaulté   ©r1   rq   r+   ÚyrG   c                ó   — g | ]
}|d          ‘ŒS rl   © rs   s     r   rp   z,KeyPoints.from_inference.<locals>.<listcomp>Õ  s   € ÐRÐRÐR¸
˜Z¨
Ô3ÐRÐRÐRr    c                ó   — g | ]
}|d          ‘ŒS )Úclassr}   rs   s     r   rp   z,KeyPoints.from_inference.<locals>.<listcomp>Ö  s   € ÐRÐRÐR¸
 
¨7Ô 3ÐRÐRÐRr    ©r=   rA   r?   rF   )r(   r6   rK   Úhasattrre   rh   rm   Úemptyrv   r   ÚzerosrX   Úfloat32rr   Úzipr/   r
   Úastyper-   )Úclsrc   ri   ÚslotsÚkeypoint_countr=   rG   Úobject_indexrt   rw   Úslotro   r?   Úclass_namesrF   s                  r   Úfrom_inferencezKeyPoints.from_inference€  sS  € õV Ð&­Ñ-Ô-ð 	ÝðYñô ð õ
 Ð# VÑ,Ô,ð 	7Ø/×4Ò4À$ÐQUÐ4ÑVÔVÐÐÝÐ% vÑ.Ô.ð 	7Ø/×4Ò4Ñ6Ô6ÐØ×#Ò# MÑ2Ô2ð 	Ø—9’9‘;”;Ðà& }Ô5ˆð
ð 
ð
 *ð
ñ 
ô 
ˆõ ØMÐM°uÐMÑMÔMØð
ñ 
ô 
ˆõ
 ŒX•s˜;Ñ'Ô'¨¸Ð;Å2Ä:ÐNÑNÔNˆÝ”X�s ;Ñ/Ô/°Ð@ÍÌ
ÐSÑSÔSˆ
Ý8AÝ�˜UÑ#Ô#ñ9
ô 9
ð 	Hð 	HÑ4ˆLÑ4˜: |õ #& l°J¸{Ô4KÑ"LÔ"Lð Hð H‘��hØ*2°3¬-¸À#¼Ð)G��< Ð%Ñ&Ø19¸,Ô1G�
˜<¨Ð-Ñ.Ð.ðHõ ”8ÐRÐRÀkÐRÑRÔRÑSÔSˆÝ”hÐRÐRÀkÐRÑRÔRÑSÔSˆÝ$9¸;Ð#GˆàˆsØØ *Ø—_’_¥SÑ)Ô)Øð	
ñ 
ô 
ð 	
r    Úmediapipe_resultsÚresolution_whútuple[int, int]c                óD  — t          |d¦  «        rC|j        }t          |j        t          ¦  «        s!|j        €g }nkd„ |j        j        D ¦   «         g}nSt          |d¦  «        r|j        }n;t          |d¦  «        r|j        €g }n!d„ |j        D ¦   «         }nt          d¦  «        ‚t          |¦  «        dk    r|  	                    ¦   «         S g }g }|D ]„}g }g }|D ]Q}	|	j
        |d         z  |	j        |d	         z  g}
|                     |
¦  «         |                     |	j        ¦  «         ŒR|                     |¦  «         |                     |¦  «         Œ… | t          j        |t          j        ¬
¦  «        t          j        |t          j        ¬
¦  «        ¬¦  «        S )aÌ  
        Creates a `sv.KeyPoints` instance from a
        [MediaPipe](https://github.com/google-ai-edge/mediapipe)
        pose landmark detection inference result.

        Args:
            mediapipe_results: The output results from Mediapipe. It supports pose
                and face landmarks from `PoseLandmarker`, `FaceLandmarker` and the
                legacy ones from `Pose` and `FaceMesh`.
            resolution_wh: A tuple of the form `(width, height)` representing the
                resolution of the frame.

        Returns:
            A `sv.KeyPoints` object containing the keypoint coordinates and
                confidences of each keypoint.

        !!! tip
            Before you start, download model bundles from the
            [MediaPipe website](https://ai.google.dev/edge/mediapipe/solutions/vision/pose_landmarker/index#models).

        Examples:
            ```python
            from supervision import _cv2 as cv2
            import mediapipe as mp
            import supervision as sv

            image = cv2.imread("<SOURCE_IMAGE_PATH>")
            image_height, image_width, _ = image.shape
            mediapipe_image = mp.Image(
                image_format=mp.ImageFormat.SRGB,
                data=cv2.cvtColor(image, cv2.COLOR_BGR2RGB))

            options = mp.tasks.vision.PoseLandmarkerOptions(
                base_options=mp.tasks.BaseOptions(
                    model_asset_path="pose_landmarker_heavy.task"
                ),
                running_mode=mp.tasks.vision.RunningMode.IMAGE,
                num_poses=2)

            PoseLandmarker = mp.tasks.vision.PoseLandmarker
            with PoseLandmarker.create_from_options(options) as landmarker:
                pose_landmarker_result = landmarker.detect(mediapipe_image)

            key_points = sv.KeyPoints.from_mediapipe(
                pose_landmarker_result, (image_width, image_height))
            ```

            ```python
            from supervision import _cv2 as cv2
            import mediapipe as mp
            import supervision as sv

            image = cv2.imread("<SOURCE_IMAGE_PATH>")
            image_height, image_width, _ = image.shape
            mediapipe_image = mp.Image(
                image_format=mp.ImageFormat.SRGB,
                data=cv2.cvtColor(image, cv2.COLOR_BGR2RGB))

            options = mp.tasks.vision.FaceLandmarkerOptions(
                base_options=mp.tasks.BaseOptions(
                    model_asset_path="face_landmarker.task"
                ),
                output_face_blendshapes=True,
                output_facial_transformation_matrixes=True,
                num_faces=2)

            FaceLandmarker = mp.tasks.vision.FaceLandmarker
            with FaceLandmarker.create_from_options(options) as landmarker:
                face_landmarker_result = landmarker.detect(mediapipe_image)

            key_points = sv.KeyPoints.from_mediapipe(
                face_landmarker_result, (image_width, image_height))
            ```

        Úpose_landmarksNc                ó   — g | ]}|‘ŒS r}   r}   )r*   Úlandmarks     r   rp   z,KeyPoints.from_mediapipe.<locals>.<listcomp>6  s(   € ð ð ð à (ð %ðð ð r    Úface_landmarksÚmulti_face_landmarksc                ó   — g | ]	}|j         ‘Œ
S r}   )r”   )r*   Úface_landmarks     r   rp   z,KeyPoints.from_mediapipe.<locals>.<listcomp>A  s+   € ð ð ð à%ð "Ô*ðð ð r    zsUnsupported MediaPipe result type. Expected an object with pose_landmarks, face_landmarks, or multi_face_landmarks.r   r%   rz   )r=   rA   )r�   r’   r(   r6   r”   r•   r–   rK   rX   r‚   r+   r{   ÚappendÚ
visibilityr   r/   r„   )r‡   rŽ   r�   Úresultsr=   rG   ÚposeÚprediction_xyÚprediction_confidencer”   Úkeypoint_xys              r   Úfrom_mediapipezKeyPoints.from_mediapipeà  s  € õ^ Ð$Ð&6Ñ7Ô7ð 	Ø'Ô6ˆGÝÐ/Ô>ÅÑEÔEð 	Ø$Ô3Ð;Ø �G�Gðð à,=Ô,LÔ,Uðñ ô ð�Gøõ Ð&Ð(8Ñ9Ô9ð 	Ø'Ô6ˆGˆGÝÐ&Ð(>Ñ?Ô?ð 	Ø Ô5Ð=Ø��ðð à):Ô)Oðñ ô ��õ ðKñô ð õ
 ˆw‰<Œ<˜1ÒÐØ—9’9‘;”;ÐàˆØˆ
Øð 	5ð 	5ˆDØˆMØ$&Ð!Ø ð Bð B�à”J ¨qÔ!1Ñ1Ø”J ¨qÔ!1Ñ1ð�ð ×$Ò$ [Ñ1Ô1Ð1Ø%×,Ò,¨XÔ-@ÑAÔAÐAÐAà�IŠI�mÑ$Ô$Ð$Ø×ÒÐ3Ñ4Ô4Ð4Ð4àˆsÝŒx˜¥"¤*Ð-Ñ-Ô-Ý "¤¨½2¼:Ð FÑ FÔ Fð
ñ 
ô 
ð 	
r    Úultralytics_resultsc                ó<  ‡— ‰j         j                             ¦   «         dk    r|                      ¦   «         S ‰j         j                             ¦   «                              ¦   «         }‰j        j                             ¦   «                              ¦   «                              t          ¦  «        }t          j        ˆfd„|D ¦   «         ¦  «        }‰j         j        }|€dn%|                     ¦   «                              ¦   «         }t          |i} | ||||¬¦  «        S )a  
        Creates a `sv.KeyPoints` instance from a
        [YOLOv8](https://github.com/ultralytics/ultralytics) pose inference result.

        Args:
            ultralytics_results: The output Results instance from YOLOv8.

        Returns:
            A `sv.KeyPoints` object containing the keypoint coordinates, class IDs,
                and class names, and confidences of each keypoint.

        Examples:
            ```python
            from supervision import _cv2 as cv2
            import supervision as sv
            from ultralytics import YOLO

            image = cv2.imread("<SOURCE_IMAGE_PATH>")
            model = YOLO('yolov8s-pose.pt')

            result = model(image)[0]
            key_points = sv.KeyPoints.from_ultralytics(result)
            ```
        r   c                ó*   •— g | ]}‰j         |         ‘ŒS r}   )Únames)r*   r!   r¡   s     €r   rp   z.KeyPoints.from_ultralytics.<locals>.<listcomp>ƒ  s"   ø€ ÐOÐOÐOÀÐ 3Ô 9¸!Ô <ÐOÐOÐOr    N)r=   r?   rA   rF   )rq   r=   Únumelr‚   ÚcpuÚnumpyÚboxesr‡   r†   r-   r   r/   Úconfr
   )r‡   r¡   r=   r?   rŒ   Úkeypoints_confrG   rF   s    `      r   Úfrom_ultralyticszKeyPoints.from_ultralyticsd  sÿ   ø€ ð4 Ô(Ô+×1Ò1Ñ3Ô3°qÒ8Ð8Ø—9’9‘;”;Ðà Ô*Ô-×1Ò1Ñ3Ô3×9Ò9Ñ;Ô;ˆØ&Ô,Ô0×4Ò4Ñ6Ô6×<Ò<Ñ>Ô>×EÒEÅcÑJÔJˆÝ”hÐOÐOÐOÐOÀhÐOÑOÔOÑPÔPˆð -Ô6Ô;ˆØ+Ð3�T�T¸×9KÒ9KÑ9MÔ9M×9SÒ9SÑ9UÔ9Uˆ
Ý$9¸;Ð#GˆØˆs�b 8ÀÐRVÐWÑWÔWÐWr    Úyolo_nas_resultsc                ó®  — t          |j        j        ¦  «        dk    r|                      ¦   «         S |j        j        dd…dd…dd…f         }|j        j        dd…dd…df         }t	          |j        d¦  «        r|j        j        }nd}i }|�:|j        �3g }|D ]$}|j        |         }|                     |¦  «         Œ%||t          <    | ||||¬¦  «        S )aã  
        Create a `sv.KeyPoints` instance from a [YOLO-NAS](https://github.com/Deci-AI/super-gradients/blob/master/YOLONAS-POSE.md)
        pose inference results.

        Args:
            yolo_nas_results: The output object from YOLO NAS.

        Returns:
            A `sv.KeyPoints` object containing the keypoint coordinates, class IDs,
                and class names, and confidences of each keypoint.

        Examples:
            ```python
            from supervision import _cv2 as cv2
            import torch
            import supervision as sv
            import super_gradients

            image = cv2.imread("<SOURCE_IMAGE_PATH>")

            device = "cuda" if torch.cuda.is_available() else "cpu"
            model = super_gradients.training.models.get(
                "yolo_nas_pose_s", pretrained_weights="coco_pose").to(device)

            results = model.predict(image, conf=0.1)
            key_points = sv.KeyPoints.from_yolo_nas(results)
            ```
        r   Nry   Úlabelsr€   )	rX   rt   Úposesr‚   r�   r®   rŒ   r™   r
   )	r‡   r¬   r=   rG   r?   rF   rŒ   Úc_idÚnames	            r   Úfrom_yolo_naszKeyPoints.from_yolo_nasŒ  s  € õ< ÐÔ*Ô0Ñ1Ô1°QÒ6Ð6Ø—9’9‘;”;ÐàÔ(Ô.¨q¨q¨q°!°!°!°R°a°R¨xÔ8ˆØ%Ô0Ô6°q°q°q¸!¸!¸!¸Q°wÔ?ˆ
õ Ð#Ô.°Ñ9Ô9ð 	Ø'Ô2Ô9ˆHˆHàˆHà#%ˆØÐÐ$4Ô$@Ð$LØˆKØ ð )ð )�Ø'Ô3°DÔ9�Ø×"Ò" 4Ñ(Ô(Ð(Ð(Ø*5ˆDÕ&Ñ'àˆsØØ *ØØð	
ñ 
ô 
ð 	
r    Údetectron2_resultsc                ó’  — t          |d         d¦  «        �r|d         j                             ¦   «                              ¦   «         j        dk    r|                      ¦   «         S  | |d         j                             ¦   «                              ¦   «         dd…dd…dd…f         |d         j                             ¦   «                              ¦   «         dd…dd…df         |d         j                             ¦   «                              ¦   «                              t          ¦  «        ¬¦  «        S |                      ¦   «         S )a  
        Create a `sv.KeyPoints` object from the
        [Detectron2](https://github.com/facebookresearch/detectron2) inference result.

        Args:
            detectron2_results: The output of a
                Detectron2 model containing instances with prediction data.

        Returns:
            A `sv.KeyPoints` object containing the keypoint coordinates, class IDs,
                and class names, and confidences of each keypoint.

        Examples:
            ```python
            from supervision import _cv2 as cv2
            import supervision as sv
            from detectron2.engine import DefaultPredictor
            from detectron2.config import get_cfg


            image = cv2.imread("<SOURCE_IMAGE_PATH>")
            cfg = get_cfg()
            cfg.merge_from_file("<CONFIG_PATH>")
            cfg.MODEL.WEIGHTS = "<WEIGHTS_PATH>"
            predictor = DefaultPredictor(cfg)

            result = predictor(image)
            keypoints = sv.KeyPoints.from_detectron2(result)
            ```
        Ú	instancesÚpred_keypointsr   Nry   ©r=   rA   r?   )	r�   r¶   r¦   r§   Úsizer‚   Úpred_classesr†   r-   )r‡   r³   s     r   Úfrom_detectron2zKeyPoints.from_detectron2Æ  s  € õB Ð% kÔ2Ð4DÑEÔEñ 	Ø! +Ô.Ô=×AÒAÑCÔC×IÒIÑKÔKÔPÐTUÒUÐUØ—y’y‘{”{Ð"Ø�3Ø% kÔ2Ü§¢¡¤ß’‘”˜˜˜˜A˜A˜A˜r ˜r˜ô#ð %7°{Ô$CÜ§¢¡¤ß’‘”˜˜˜˜A˜A˜A˜q˜ô%"ð ,¨KÔ8ÜŸcšc™eœeß’‘”ß’�‘”ðñ ô ð ð —9’9‘;”;Ðr    Útransformers_resultsc           	     óT  — d|d         v �r
|d         d                               ¦   «                              ¦   «         j        dk    r|                      ¦   «         S d„ |D ¦   «         }t	          |Ž \  }} | t          j        |¦  «                             t
          j        ¦  «        t          j        |¦  «                             t
          j        ¦  «        t          j	        t          |¦  «        ¦  «                             t          ¦  «        ¬¦  «        S |                      ¦   «         S )ap	  
        Create a `sv.KeyPoints` object from the
        [Transformers](https://github.com/huggingface/transformers) inference result.

        Args:
            transformers_results: The output of a
                Transformers model containing instances with prediction data.

        Returns:
            A `sv.KeyPoints` object containing the keypoint coordinates, class IDs,
                and class names, and confidences of each keypoint.

        Examples:
            ```python
            from PIL import Image
            import requests
            import supervision as sv
            import torch
            from transformers import (
                AutoProcessor,
                RTDetrForObjectDetection,
                VitPoseForPoseEstimation,
            )

            device = "cuda" if torch.cuda.is_available() else "cpu"
            image = Image.open("<SOURCE_IMAGE_PATH>")

            DETECTION_MODEL_ID = "PekingU/rtdetr_r50vd_coco_o365"

            detection_processor = AutoProcessor.from_pretrained(DETECTION_MODEL_ID, use_fast=True)
            detection_model = RTDetrForObjectDetection.from_pretrained(DETECTION_MODEL_ID, device_map=device)

            inputs = detection_processor(images=frame, return_tensors="pt").to(device)

            with torch.no_grad():
                outputs = detection_model(**inputs)

            target_size = torch.tensor([(frame.height, frame.width)])
            results = detection_processor.post_process_object_detection(
                outputs, target_sizes=target_size, threshold=0.3)

            detections = sv.Detections.from_transformers(results[0])
            boxes = sv.xyxy_to_xywh(detections[detections.class_id == 0].xyxy)

            POSE_ESTIMATION_MODEL_ID = "usyd-community/vitpose-base-simple"

            pose_estimation_processor = AutoProcessor.from_pretrained(POSE_ESTIMATION_MODEL_ID)
            pose_estimation_model = VitPoseForPoseEstimation.from_pretrained(
                POSE_ESTIMATION_MODEL_ID, device_map=device)

            inputs = pose_estimation_processor(frame, boxes=[boxes], return_tensors="pt").to(device)

            with torch.no_grad():
                outputs = pose_estimation_model(**inputs)

            results = pose_estimation_processor.post_process_pose_estimation(outputs, boxes=[boxes])
            key_point = sv.KeyPoints.from_transformers(results[0])
            ```

        rq   r   c                ó¾   — g | ]Z}|d                                ¦   «                              ¦   «         |d                               ¦   «                              ¦   «         f‘Œ[S )rq   Úscores)r¦   r§   )r*   Úresults     r   rp   z/KeyPoints.from_transformers.<locals>.<listcomp><  sl   € ð ð ð ð
 ð ˜;Ô'×+Ò+Ñ-Ô-×3Ò3Ñ5Ô5Ø˜8Ô$×(Ò(Ñ*Ô*×0Ò0Ñ2Ô2ððð ð r    r·   )r¦   r§   r¸   r‚   r…   r   Ústackr†   r„   ÚarangerX   r-   )r‡   r»   Úresult_datar=   r¾   s        r   Úfrom_transformerszKeyPoints.from_transformersù  s  € ð~ Ð.¨qÔ1Ð1Ñ1Ø# AÔ& {Ô3×7Ò7Ñ9Ô9×?Ò?ÑAÔAÔFÈ!ÒKÐKØ—y’y‘{”{Ð"ðð ð
 3ðñ ô ˆKõ ˜kÐ*‰JˆB�à�3Ý”8˜B‘<”<×&Ò&¥r¤zÑ2Ô2Ý$&¤H¨VÑ$4Ô$4×$;Ò$;½B¼JÑ$GÔ$GÝœ¥3 r¡7¤7Ñ+Ô+×2Ò2µ3Ñ7Ô7ðñ ô ð ð —9’9‘;”;Ðr    Úmaskúnpt.NDArray[np.bool_]c                ó^  — t          | j        ¦  «        }|j        d         |k    r!t          d|j        d         › d|› d�¦  «        ‚|j        d         | j        j        d         k    r1t          d|j        d         › d| j        j        d         › d�¦  «        ‚t	          j        |d¬¦  «        }|dk    rBt	          j        ||d         k    ¦  «        s$t          d	|                     ¦   «         › �¦  «        ‚|dk    rt          |d         ¦  «        nd}t	          j	        ||| j        j        d
         f| j        j
        ¬¦  «        }d}| j        �Jt          t          j        t          j                 t	          j	        ||f| j        j
        ¬¦  «        ¦  «        }d}| j        �t	          j	        ||ft"          ¬¦  «        }t%          |¦  «        D ]d}t	          j        ||         ¦  «        }	| j        ||	f         ||<   |�| j        �| j        ||	f         ||<   |�| j        �| j        ||	f         ||<   Œed}
| j        �| j                             ¦   «         }
d}| j        �| j                             ¦   «         }t/          | j        t3          d¦  «        ¦  «        }t5          |||
|||¬¦  «        S )u×  Filter keypoints using a 2D boolean mask of shape `(n, m)`.

        This method selects the **same set of keypoints from every object**, so
        every row of `mask` must contain the same number of `True` values.  The
        result is a new `KeyPoints` whose keypoint count is that uniform `k`.

        This is suitable for use cases such as *"keep only the left-side joints for
        all persons"* â€” where the selected joint indices are identical across objects.

        It is **not** suitable for per-object confidence filtering
        (`kp[kp.confidence > 0.5]`) when the threshold yields a different number of
        passing keypoints per object, because NumPy cannot represent a ragged
        `(n, ?, 2)` array.  For that pattern either process objects individually or
        zero out low-confidence entries in-place via `kp.confidence`.

        For the single-object case (`n == 1`) any boolean mask always satisfies the
        uniform-count requirement, so `kp[kp.confidence > 0.5]` works as expected.

        Args:
            mask: A boolean array of shape `(n, m)` where `n` is the number of
                objects and `m` is the number of keypoints per object.  Every row
                must select the same number of keypoints so that the result can be
                stored in a uniform `(n, k, ...)` array.

        Returns:
            A new `KeyPoints` instance containing only the keypoints selected by
            the mask for each object.

        Raises:
            ValueError: If `mask.shape[0]` does not match the number of objects, if
                `mask.shape[1]` does not match the number of keypoints, or if
                different rows of the mask select different numbers of `True` values.
        r   z2D boolean mask row count z does not match object count ú.r%   z2D boolean mask column count z does not match keypoint count ©Úaxisz®Cannot filter keypoints with a 2D boolean mask where rows have different numbers of True values. All objects must select the same number of keypoints. Got counts per object: ry   rz   N©r=   rA   rB   rD   r?   rF   )rX   r=   ÚshaperK   r   Úsumr7   Útolistr-   rƒ   r1   rA   r	   ÚnptÚNDArrayr„   rD   r   r\   r2   rB   Úcopyr?   r   rF   Úslicer;   )rM   rÄ   ÚnÚcountsÚkÚxy_selectedÚkeypoint_confidence_selectedÚvisible_selectedÚrowÚrow_indicesÚdetection_confidence_selectedÚclass_id_selectedÚdata_selecteds                r   Ú_get_by_2d_bool_maskzKeyPoints._get_by_2d_bool_maskN  sî  € õD �”‰LŒLˆØŒ:�aŒ=˜AÒÐÝð%¨T¬Z¸¬]ð %ð %Ø !ð%ð %ð %ñô ð ð Œ:�aŒ=˜DœGœM¨!Ô,Ò,Ð,Ýð6°´
¸1´ð 6ð 6Ø"&¤'¤-°Ô"2ð6ð 6ð 6ñô ð õ ”˜ 1Ð%Ñ%Ô%ˆØˆqŠ5ˆ5�œ ¨&°¬)Ò 3Ñ4Ô4ˆ5Ýð<ð +1¯-ª-©/¬/ð<ð <ñô ð ð   !še˜e�C��q”	‰NŒNˆN¨ˆÝ”h  1 d¤g¤m°AÔ&6Ð7¸t¼w¼}ÐMÑMÔMˆØGKÐ$ØÔ#Ð/Ý+/Ý”�BœJÔ'Ý”˜!˜Q˜ tÔ'?Ô'EÐFÑFÔFñ,ô ,Ð(ð :>ÐØŒ<Ð#Ý!œx¨¨A¨µdÐ;Ñ;Ô;ÐÝ˜‘8”8ð 	Gð 	GˆCÝœ.¨¨c¬Ñ3Ô3ˆKØ#œw s¨KÐ'7Ô8ˆK˜Ñà,Ð8ØÔ,Ð8à48Ô4LØ˜Ð$ô5Ð,¨SÑ1ð  Ð+°´Ð0HØ(,¬°S¸+Ð5EÔ(FÐ  Ñ%øØ(,Ð%ØÔ$Ð0Ø,0Ô,E×,JÒ,JÑ,LÔ,LÐ)à ÐØŒ=Ð$Ø $¤× 2Ò 2Ñ 4Ô 4Ðå% d¤iµ°t±´Ñ=Ô=ˆåØØ <Ø!>Ø$Ø&Øð
ñ 
ô 
ð 	
r    ÚkeyÚstrú_DetectionDataValueType | Nonec                ó6   — | j                              |¦  «        S )aD  Get a value from the keypoint data dictionary.

        Args:
            key: Data field name.

        Returns:
            The stored data value, or `None` when the key is absent.

        Example:
            ```pycon
            >>> import numpy as np
            >>> from supervision import KeyPoints
            >>> key_points = KeyPoints(
            ...     xy=np.array([[[0, 0]]]),
            ...     data={"class_name": np.array(["person"])},
            ... )
            >>> key_points.get_data("class_name").tolist()
            ['person']

            ```
        )rF   rm   )rM   rÞ   s     r   Úget_datazKeyPoints.get_data­  s   € ð, Œy�}Š}˜SÑ!Ô!Ð!r    ÚindexúIndex1D | Index2Dc                óŠ	  — t          |t          j        ¦  «        rX|j        dk    rM|j        t
          k    r=|                      t          t          j	        t          j
                 |¦  «        ¦  «        S t          |t          ¦  «        s|t          d¦  «        f}|\  }}t          |t          ¦  «        r|g}t          |t          ¦  «        r/|r-t          d„ |D ¦   «         ¦  «        rt          j        |¦  «        }t          |t          ¦  «        r/|r-t          d„ |D ¦   «         ¦  «        rt          j        |¦  «        }t          |t          j        ¦  «        r$|j        t
          k    rt          j        |¦  «        }t          |t          j        ¦  «        r$|j        t
          k    rt          j        |¦  «        }|}t          |t          t          j        f¦  «        r±t          |t          t          j        f¦  «        r�t          j        |¦  «        s|t          j        |¦  «        sht          j        t          t(          |¦  «        t          t(          |¦  «        ¦  «        \  }}t          t(          |¦  «        }t          t(          |¦  «        }t+          |¦  «        }| j        ||f         }d}	| j        �| j        ||f         }	d}
| j        �| j        |         }
d}| j        �| j        ||f         }| j        �| j        |         nd}t7          | j        t          t(          |¦  «        ¦  «        }|j        dk    rI|                     ddd¦  «        }|	�|	                     dd¦  «        }	|�|                     dd¦  «        }�n,|j        dk    �r t          j        |d         ¦  «        s1t          |d         t          j        ¦  «        rR|d         j        dk    rA|t          j        df         }|	�|	t          j        df         }	|�|t          j        df         }n”t          j        |d         ¦  «        s1t          |d         t          j        ¦  «        rZ|d         j        dk    rI|dd…t          j        dd…f         }|	�|	dd…t          j        f         }	|�|dd…t          j        f         }t?          ||	|
|||¬¦  «        S )	a�  Get a subset of the KeyPoints object.

        Supports detection-level (skeleton) filtering, keypoint-level (anchor)
        filtering, and combined tuple indexing.

        Args:
            index: Index, indices, slice, or boolean mask selecting key points.

        Returns:
            A new `KeyPoints` instance containing the selected rows or anchors.

        Example:
            ```pycon
            >>> import numpy as np
            >>> from supervision import KeyPoints
            >>> key_points = KeyPoints(xy=np.array([[[0, 1]], [[2, 3]]]))
            >>> key_points.select([1]).xy.tolist()
            [[[2, 3]]]

            ```
        ry   Nc              3  ó@   K  — | ]}t          |t          ¦  «        V — Œd S r   r'   r)   s     r   r,   z#KeyPoints.select.<locals>.<genexpr>é  ó,   è è € Ð,LÐ,LÀQ­Z¸½4Ñ-@Ô-@Ð,LÐ,LÐ,LÐ,LÐ,LÐ,Lr    c              3  ó@   K  — | ]}t          |t          ¦  «        V — Œd S r   r'   r)   s     r   r,   z#KeyPoints.select.<locals>.<genexpr>ë  rç   r    r%   r   .rÊ   ) r(   r   r0   r4   r1   r   rÝ   r	   rÎ   rÏ   Úbool_ÚtuplerÑ   r-   r6   r7   r/   r2   ÚisscalarÚix_r   r8   r=   rA   rB   rD   r?   r   rF   r5   Únewaxisr;   )rM   rã   r!   ÚjÚraw_iÚi_ixÚj_ixÚrow_irÕ   rÖ   rÚ   r×   rÛ   rÜ   s                 r   ÚselectzKeyPoints.selectÅ  s�  € õ2 �e�RœZÑ(Ô(ð 	Q¨U¬Z¸1ª_¨_ÀÄÕPTÒATÐATØ×,Ò,­Tµ#´+½b¼hÔ2GÈÑ-OÔ-OÑPÔPÐPå˜%¥Ñ'Ô'ð 	)Ø�E $™KœKÐ(ˆEà‰ˆˆ1å�a�ÑÔð 	Ø�ˆAå�a�ÑÔð 	 1ð 	­Ð,LÐ,LÈ!Ð,LÑ,LÔ,LÑ)LÔ)Lð 	Ý”˜‘”ˆAÝ�a�ÑÔð 	 1ð 	­Ð,LÐ,LÈ!Ð,LÑ,LÔ,LÑ)LÔ)Lð 	Ý”˜‘”ˆAå�a�œÑ$Ô$ð 	"¨¬µDª¨Ý”˜qÑ!Ô!ˆAÝ�a�œÑ$Ô$ð 	"¨¬µDª¨Ý”˜qÑ!Ô!ˆAàˆõ �q�4¥¤Ð,Ñ-Ô-ð	 å˜1�t¥R¤ZÐ0Ñ1Ô1ð	 õ ”K ‘N”Nð	 õ ”K ‘N”Nð		 õ œ¥¥S¨!¡¤­dµ3¸©l¬lÑ;Ô;‰JˆD�$Ý•S˜$‘”ˆAÝ•S˜$‘”ˆAå$ UÑ+Ô+ˆà”g˜a ˜d”mˆà'+Ð$ØÔ#Ð/Ø+/Ô+CÀAÀqÀDÔ+IÐ(à(,Ð%ØÔ$Ð0Ø,0Ô,EÀeÔ,LÐ)àÐØŒ<Ð#Ø#œ|¨A¨q¨DÔ1Ðà48´MÐ4M˜DœM¨%Ô0Ð0ÐSWÐå% d¤iµµc¸5Ñ1AÔ1AÑBÔBˆàÔ˜qÒ Ð Ø%×-Ò-¨a°°AÑ6Ô6ˆKØ+Ð7Ø/K×/SÒ/SØ�qñ0ô 0Ð,ð  Ð+Ø#3×#;Ò#;¸A¸qÑ#AÔ#AÐ ùØÔ Ò"Ñ"ÝŒ{˜5 œ8Ñ$Ô$ð GÝ˜5 œ8¥R¤ZÑ0Ô0ðGØ5:¸1´X´]ÀaÒ5GÐ5Gà)­"¬*°c¨/Ô:�Ø/Ð;Ø3OÝœ
 C˜ô4Ð0ð $Ð/Ø'7½¼
ÀC¸Ô'HÐ$øÝ”˜U 1œXÑ&Ô&ð 	GÝ˜5 œ8¥R¤ZÑ0Ô0ð	GØ5:¸1´X´]ÀaÒ5GÐ5Gà)¨!¨!¨!­R¬Z¸¸¸Ð*:Ô;�Ø/Ð;Ø3OØ˜˜�2œ:˜ô4Ð0ð $Ð/Ø'7¸¸¸½2¼:¸Ô'FÐ$åØØ <Ø!>Ø$Ø&Øð
ñ 
ô 
ð 	
r    úIndex1D | Index2D | strú6KeyPoints | npt.NDArray[np.generic] | list[Any] | Nonec                ó€   — t          |t          ¦  «        r|                      |¦  «        S |                      |¦  «        S )a	  
        Get a subset of the KeyPoints object or access an item from its data field.

        Supports detection-level (skeleton) filtering, keypoint-level (anchor)
        filtering, combined tuple indexing, and data field access by string key.

        Args:
            index: The index, indices, or key to access a subset of the KeyPoints
                or an item from the data.

        Returns:
            A subset of the KeyPoints object or an item from the data field.

        Examples:
            ```python
            import supervision as sv

            key_points = sv.KeyPoints(...)

            # detection-level filtering (returns KeyPoints)
            high_conf = key_points[key_points.detection_confidence > 0.5]
            class_0 = key_points[key_points.class_id == 0]

            # keypoint-level filtering (returns KeyPoints)
            visible = key_points[key_points.keypoint_confidence > 0.3]

            # indexing
            first = key_points[0]
            first_two = key_points[0:2]
            subset = key_points[[0, 2]]

            # anchor selection (uniform across all skeletons)
            nose_and_eyes = key_points[:, [0, 1, 2]]

            # data field access
            class_names = key_points['class_name']
            ```
        )r(   rß   râ   ró   )rM   rã   s     r   Ú__getitem__zKeyPoints.__getitem__:  s;   € õT �e�SÑ!Ô!ð 	(Ø—=’= Ñ'Ô'Ð'Ø�{Š{˜5Ñ!Ô!Ð!r    ú#npt.NDArray[np.generic] | list[Any]c                óÌ   — t          |t          j        t          f¦  «        st	          d¦  «        ‚t          |t          ¦  «        rt          j        |¦  «        }|| j        |<   dS )að  
        Set a value in the data dictionary of the `sv.KeyPoints` object.

        Args:
            key: The key in the data dictionary to set.
            value: The value to set for the key.

        Examples:
            ```python
            from supervision import _cv2 as cv2
            import supervision as sv
            from ultralytics import YOLO

            image = cv2.imread("<SOURCE_IMAGE_PATH>")
            model = YOLO('yolov8s.pt')

            result = model(image)[0]
            key_points = sv.KeyPoints.from_ultralytics(result)

            key_points['class_name'] = [
                 model.model.names[class_id]
                 for class_id
                 in key_points.class_id
             ]
            ```
        z$Value must be a np.ndarray or a listN)r(   r   r0   r6   Ú	TypeErrorr/   rF   )rM   rÞ   rU   s      r   Ú__setitem__zKeyPoints.__setitem__h  s[   € õ6 ˜%¥"¤*­dÐ!3Ñ4Ô4ð 	DÝÐBÑCÔCÐCå�e�TÑ"Ô"ð 	$Ý”H˜U‘O”OˆEàˆŒ	�#‰ˆˆr    c                óV   —  | t          j        dt           j        ¬¦  «        ¬¦  «        S )aF  
        Create an empty KeyPoints object with no key points.

        Returns:
            An empty `sv.KeyPoints` object.

        Examples:
            ```pycon
            >>> import supervision as sv
            >>> key_points = sv.KeyPoints.empty()
            >>> len(key_points)
            0

            ```
        )r   r   ry   rz   ©r=   )r   r‚   r„   )r‡   s    r   r‚   zKeyPoints.empty‹  s(   € ð" ˆs•b”h˜yµ´
Ð;Ñ;Ô;Ð<Ñ<Ô<Ð<r    c                ó(   — t          | ¦  «        dk    S )ai  
        Returns `True` if the `KeyPoints` object is considered empty.

        Returns:
            `True` if the object is empty, `False` otherwise.

        Example:
            ```pycon
            >>> import supervision as sv
            >>> key_points = sv.KeyPoints.empty()
            >>> key_points.is_empty()
            True

            ```
        r   )rX   rQ   s    r   Úis_emptyzKeyPoints.is_emptyž  s   € õ  �4‰yŒy˜AŠ~Ðr    Úkey_points_listúlist[KeyPoints]c           	     ó"  ‡— d„ ‰D ¦   «         Št          ‰¦  «        dk    r|                      ¦   «         S ‰D ]5}t          |j        |j        |j        |j        |j        |j        ¬¦  «         Œ6d„ ‰D ¦   «         }t          |¦  «        dk    r t          dt          |¦  «        › d�¦  «        ‚d„ ‰D ¦   «         }t          |¦  «        dk    r t          d	t          |¦  «        › d�¦  «        ‚t          j        d
„ ‰D ¦   «         ¦  «        }dˆfd„}t          t          j        t          j                 dz   |d¦  «        ¦  «        }t          t          j        t          j                 dz   |d¦  «        ¦  «        }t          t          j        t          j                 dz   |d¦  «        ¦  «        }	t          t          j        t          j                 dz   |d¦  «        ¦  «        }
t'          d„ ‰D ¦   «         ¦  «        } | ||||	|
|¬¦  «        S )aâ  
        Merge a list of KeyPoints objects into a single KeyPoints object.

        This method takes a list of KeyPoints objects and combines their
        respective fields (`xy`, `class_id`, `keypoint_confidence`,
        `detection_confidence`, and `visible`) into a single KeyPoints object.
        The `data` dictionaries are merged key-wise, following the same rules
        as `Detections.merge`.

        For example, if merging KeyPoints with 2 and 3 skeletons, this method
        will return a KeyPoints with 5 skeletons (5 entries in `xy`, etc).

        !!! Note

            When merging, empty `KeyPoints` objects are ignored.

        Args:
            key_points_list: A list of KeyPoints objects to merge.

        Returns:
            A single KeyPoints object containing the merged data from the input list.

        Raises:
            ValueError: If the non-empty inputs do not share the same number
                of keypoints per skeleton.
            ValueError: If the non-empty inputs do not share the same
                coordinate depth (`xy.shape[2]`) per skeleton.
            ValueError: If some inputs have a field set (`class_id`,
                `keypoint_confidence`, `detection_confidence`, or `visible`)
                and others do not.
            ValueError: If the input `data` dictionaries do not share the
                same keys.

        Example:
            ```pycon
            >>> import numpy as np
            >>> import supervision as sv
            >>> key_points_1 = sv.KeyPoints(
            ...     xy=np.array([[[10, 10], [20, 20]]], dtype=np.float32),
            ...     class_id=np.array([0]),
            ...     data={'class_name': np.array(['person'])},
            ... )
            >>> key_points_2 = sv.KeyPoints(
            ...     xy=np.array([[[30, 30], [40, 40]]], dtype=np.float32),
            ...     class_id=np.array([1]),
            ...     data={'class_name': np.array(['dog'])},
            ... )
            >>> merged = sv.KeyPoints.merge([key_points_1, key_points_2])
            >>> len(merged)
            2
            >>> merged.class_id
            array([0, 1])
            >>> merged.data['class_name']
            array(['person', 'dog'], dtype='<U6')

            ```
        c                ó:   — g | ]}|                      ¦   «         °|‘ŒS r}   )rÿ   ©r*   Ú
key_pointss     r   rp   z#KeyPoints.merge.<locals>.<listcomp>ë  s9   € ð 
ð 
ð 
Ø%À
×@SÒ@SÑ@UÔ@Uð
Øð
ð 
ð 
r    r   rP   c                ó2   — h | ]}|j         j        d          ’ŒS )r%   ©r=   rË   r  s     r   ú	<setcomp>z"KeyPoints.merge.<locals>.<setcomp>ü  ó"   € ÐTÐTÐT°j˜:œ=Ô.¨qÔ1ÐTÐTÐTr    r%   z[All KeyPoints must have the same number of keypoints per skeleton to be merged; got counts rÇ   c                ó2   — h | ]}|j         j        d          ’ŒS )ry   r  r  s     r   r  z"KeyPoints.merge.<locals>.<setcomp>  r	  r    zXAll KeyPoints must have the same coordinate depth per skeleton to be merged; got depths c                ó   — g | ]	}|j         ‘Œ
S r}   rý   r  s     r   rp   z#KeyPoints.merge.<locals>.<listcomp>
  s   € ÐHÐHÐH¨*˜
œÐHÐHÐHr    r±   rß   r   r   c                ó,  •‡ — ˆ fd„‰D ¦   «         }t          d„ |D ¦   «         ¦  «        rd S t          d„ |D ¦   «         ¦  «        rt          d‰ › d�¦  «        ‚t          t          j        t          j                 t          j        |d¬¦  «        ¦  «        S )Nc                ó0   •— g | ]}t          |‰¦  «        ‘ŒS r}   )Úgetattr)r*   r  r±   s     €r   rp   z:KeyPoints.merge.<locals>.stack_or_none.<locals>.<listcomp>  s#   ø€ ÐRÐRÐR°J•g˜j¨$Ñ/Ô/ÐRÐRÐRr    c              3  ó   K  — | ]}|d u V — Œ	d S r   r}   ©r*   rU   s     r   r,   z9KeyPoints.merge.<locals>.stack_or_none.<locals>.<genexpr>  ó&   è è € Ð5Ð5 U�5˜D�=Ð5Ð5Ð5Ð5Ð5Ð5r    c              3  ó   K  — | ]}|d u V — Œ	d S r   r}   r  s     r   r,   z9KeyPoints.merge.<locals>.stack_or_none.<locals>.<genexpr>  r  r    zAll or none of the 'z' fields must be Noner   rÈ   )	r7   ÚanyrK   r	   rÎ   rÏ   r   ÚgenericÚconcatenate)r±   Úvaluesr   s   ` €r   Ústack_or_nonez&KeyPoints.merge.<locals>.stack_or_none  s£   øø€ ØRÐRÐRÐRÀ/ÐRÑRÔRˆFÝÐ5Ð5¨fÐ5Ñ5Ô5Ñ5Ô5ð Ø�tÝÐ5Ð5¨fÐ5Ñ5Ô5Ñ5Ô5ð UÝ Ð!S¸Ð!SÐ!SÐ!SÑTÔTÐTÝ�œ¥B¤JÔ/µ´ÀÈQÐ1OÑ1OÔ1OÑPÔPÐPr    Nr?   rA   rB   rD   c                ó   — g | ]	}|j         ‘Œ
S r}   )rF   r  s     r   rp   z#KeyPoints.merge.<locals>.<listcomp>  s   € ÐMÐMÐM¨z˜:œ?ÐMÐMÐMr    )r=   r?   rA   rB   rD   rF   )r±   rß   r   r   )rX   r‚   r   r=   r?   rA   rB   rD   rF   rK   Úsortedr   Úvstackr	   rÎ   rÏ   Úint_r„   ré   r   )r‡   r   r  Úkeypoint_countsÚkeypoint_depthsr=   r  r?   rA   rB   rD   rF   s    `          r   ÚmergezKeyPoints.merge°  s‰  ø€ ðv
ð 
Ø)8ð
ñ 
ô 
ˆõ ˆÑÔ 1Ò$Ð$Ø—9’9‘;”;Ðà)ð 	ð 	ˆJÝ&Ø”=Ø#Ô,Ø%Ô9Ø%/Ô%DØ"Ô*Ø”_ðñ ô ð ð ð UÐTÀOÐTÑTÔTˆÝˆÑÔ !Ò#Ð#ÝðPÝ5;¸OÑ5LÔ5LðPð Pð Pñô ð ð
 UÐTÀOÐTÑTÔTˆÝˆÑÔ !Ò#Ð#ÝðPÝ5;¸OÑ5LÔ5LðPð Pð Pñô ð õ
 ŒYÐHÐH¸ÐHÑHÔHÑIÔIˆð	Qð 	Qð 	Qð 	Qð 	Qð 	Qõ �œ¥B¤GÔ,¨tÑ3°]°]À:Ñ5NÔ5NÑOÔOˆÝ"ÝŒK�œ
Ô# dÑ*¨M¨MÐ:OÑ,PÔ,Pñ
ô 
Ðõ  $ÝŒK�œ
Ô# dÑ*¨M¨MÐ:PÑ,QÔ,Qñ 
ô  
Ðõ •s”{¥2¤8Ô,¨tÑ3°]°]À9Ñ5MÔ5MÑNÔNˆåÐMÐM¸_ÐMÑMÔMÑNÔNˆàˆsØØØ 3Ø!5ØØð
ñ 
ô 
ð 	
r    g      à?FÚ	thresholdÚfloatÚclass_agnosticÚoverlap_metricr   c                óR  — t          | ¦  «        dk    r| S | j        €t          d¦  «        ‚|s| j        €t          d¦  «        ‚| j        }t          j        |dk    d¬¦  «         t          j        |¦  «                             d¬¦  «        z  }| j        �
|| j        z  }| 	                    d¬¦  «        }t          j
        t          j        ||d         t
          j        ¦  «        d¬¦  «        }t          j
        t          j        ||d	         t
          j        ¦  «        d¬¦  «        }t          j        t          j        ||d         t
          j         ¦  «        d¬¦  «        }	t          j        t          j        ||d	         t
          j         ¦  «        d¬¦  «        }
t          j        |||	|
gd¬¦  «                             t
          j        ¦  «        }d
|| <   |r0t          j        || j                             dd¦  «        g¦  «        }nst'          t(          j        t
          j                 | j        ¦  «        }t          j        || j                             dd¦  «        |                     dd¦  «        g¦  «        }t/          |||¬¦  «        }|                      |¦  «        S )ae  Performs non-max suppression on the keypoint detections. Bounding boxes are
        derived from valid keypoints of each skeleton, and standard box NMS is applied.
        A keypoint is considered valid when its coordinates are finite and not all-zero,
        and its `visible` flag is `True` (if `visible` is set). A skeleton left without
        a valid keypoint keeps a zero-area box, so it overlaps nothing and passes
        through.

        Args:
            threshold: The intersection-over-union threshold to use for
                non-maximum suppression. Must be in [0, 1]. Defaults to 0.5.
            class_agnostic: Whether to perform class-agnostic non-maximum
                suppression. If True, the class_id of each detection will be
                ignored. Defaults to False.
            overlap_metric: Metric used to compute the degree of overlap
                between pairs of bounding boxes. Defaults to
                `OverlapMetric.IOU`.

        Returns:
            A new `sv.KeyPoints` object after non-maximum suppression.

        Raises:
            ValueError: If `detection_confidence` is None.
            ValueError: If `class_agnostic` is False and `class_id`
                is None.

        Examples:
            ```python
            from supervision import _cv2 as cv2
            import supervision as sv
            from rfdetr import RFDETRKeypointPreview

            image = cv2.imread("<SOURCE_IMAGE_PATH>")
            model = RFDETRKeypointPreview()

            key_points = model.predict(image)
            key_points = key_points.with_nms(threshold=0.5)
            ```
        r   NzDKeyPoints detection_confidence must be given for NMS to be executed.zKeyPoints class_id must be given for NMS to be executed. If you intended to perform class agnostic NMS set class_agnostic=True.éÿÿÿÿrÈ   r%   ).r   ).r%   ç        )ri   Úiou_thresholdr"  )rX   rB   rK   r?   r=   r   r7   ÚisfiniterD   r  ÚminÚwhereÚinfrv   rÀ   r†   r„   Úhstackr5   r	   rÎ   rÏ   r  r   ró   )rM   r  r!  r"  r=   ÚvalidÚ	has_validÚx_minÚy_minÚx_maxÚy_maxÚxyxyri   r?   Úkeeps                  r   Úwith_nmszKeyPoints.with_nms(  sj  € õX ˆt‰9Œ9˜Š>ˆ>ØˆKàÔ$Ð,ÝØVñô ð ð ð 	 $¤-Ð"7Ýð'ñô ð ð ŒWˆõ ”˜˜aš bÐ)Ñ)Ô)Ð)­B¬K¸©O¬O×,?Ò,?ÀRÐ,?Ñ,HÔ,HÑHˆØŒ<Ð#Ø˜DœLÑ(ˆEØ—I’I 1�IÑ%Ô%ˆ	Ý”•r”x  r¨&¤zµ2´6Ñ:Ô:ÀÐCÑCÔCˆÝ”•r”x  r¨&¤zµ2´6Ñ:Ô:ÀÐCÑCÔCˆÝ”•r”x  r¨&¤zµB´F°7Ñ;Ô;À!ÐDÑDÔDˆÝ”•r”x  r¨&¤zµB´F°7Ñ;Ô;À!ÐDÑDÔDˆÝŒx˜  u¨eÐ4¸1Ð=Ñ=Ô=×DÒDÅRÄZÑPÔPˆð ˆˆiˆZÑàð 
	Ýœ) T¨4Ô+D×+LÒ+LÈRÐQRÑ+SÔ+SÐ$TÑUÔUˆKˆKå�CœK­¬Ô0°$´-Ñ@Ô@ˆHÝœ)àØÔ-×5Ò5°b¸!Ñ<Ô<Ø×$Ò$ R¨Ñ+Ô+ðñô ˆKõ 'Ø#Ø#Ø)ð
ñ 
ô 
ˆð �{Š{˜4Ñ Ô Ð r    Úselected_keypoint_indicesúIterable[int] | Noner   c                óú  — |                       ¦   «         s| j        j        d         dk    rt          j        ¦   «         S | j        }d}|�:t          j        t          |¦  «        t
          j        ¬¦  «        }|j	        dk    r|}|�|dd…|dd…f         }t          j
        |dk    d¬¦  «         t          j        |¦  «         
                    d¬¦  «        z  }| j        �||€| j        n| j        dd…|f         z  }|                     d¬¦  «        }|dd…dd…df         |dd…dd…df         }}t          j        ||t
          j        ¦  «                             d¬¦  «        }	t          j        ||t
          j        ¦  «                             d¬¦  «        }
t          j        ||t
          j         ¦  «                             d¬¦  «        }t          j        ||t
          j         ¦  «                             d¬¦  «        }t          j        |	|
||fd¬¦  «                             t
          j        ¦  «        }d|| <   | j        �%| j                             t
          j        ¦  «        }nR| j        �I| j        }|�|dd…|f         }|                     d¬¦  «                             t
          j        ¦  «        }nd}t          ||¬¦  «        }| j        |_        | j        |_        |                     |¦  «        }|S )	aí  Convert a KeyPoints object to a Detections object. This approximates the
        bounding box of the detected object by taking the bounding box that fits all key
        points. Key points that are missing (`[0, 0]`), not finite, or marked not
        visible via `visible` are left out, as in `with_nms`, and a skeleton with no
        other key point is dropped.

        Args:
            selected_keypoint_indices: The
                indices of the key points to include in the bounding box
                calculation. This helps focus on a subset of key points,
                e.g. when some are occluded. Captures all key points by
                default. An empty sequence (`[]`) is treated the same as
                `None` and selects all key points.

        Returns:
            detections: The converted detections object.

        Examples:
            ```pycon
            >>> import numpy as np
            >>> import supervision as sv
            >>> key_points = sv.KeyPoints(
            ...     xy=np.array([[[10, 20], [30, 40]]], dtype=np.float32)
            ... )
            >>> detections = key_points.as_detections()
            >>> detections.xyxy
            array([[10., 20., 30., 40.]], dtype=float32)

            ```
        r%   r   Nrz   ry   rÈ   r%  )r2  rG   )rÿ   r=   rË   r   r‚   r   Úasarrayr6   Úintpr¸   r7   r'  rD   r  r)  r*  r(  rv   rÀ   r†   r„   rB   rA   Úmeanr?   rF   ró   )rM   r5  r=   ÚindicesÚ	candidater,  r-  r+   r{   r.  r/  r0  r1  r2  rG   rA   Ú
detectionss                    r   Úas_detectionszKeyPoints.as_detectionsŠ  sÎ  € ðF �=Š=‰?Œ?ð 	&˜dœgœm¨AÔ.°!Ò3Ð3ÝÔ#Ñ%Ô%Ð%àŒWˆØ/3ˆØ$Ð0Ýœ
¥4Ð(AÑ#BÔ#BÍ"Ì'ÐRÑRÔRˆIØŒ~ Ò!Ð!Ø#�ØÐØ�A�A�A�w   �MÔ"ˆBõ ”˜˜aš aÐ(Ñ(Ô(Ð(­2¬;°r©?¬?×+>Ò+>ÀAÐ+>Ñ+FÔ+FÑFˆð Œ<Ð#Ø W _�T”\�\¸$¼,ÀqÀqÀqÈ'ÀzÔ:RÑRˆEØ—I’I 1�IÑ%Ô%ˆ	à�!�!�!�Q�Q�Q˜�'Œ{˜B˜q˜q˜q ! ! ! Q˜wœKˆ1ˆÝ”˜ ¥2¤6Ñ*Ô*×.Ò.°AÐ.Ñ6Ô6ˆÝ”˜ ¥2¤6Ñ*Ô*×.Ò.°AÐ.Ñ6Ô6ˆÝ”˜ ¥B¤F 7Ñ+Ô+×/Ò/°QÐ/Ñ7Ô7ˆÝ”˜ ¥B¤F 7Ñ+Ô+×/Ò/°QÐ/Ñ7Ô7ˆåŒx˜  u¨eÐ4¸1Ð=Ñ=Ô=×DÒDÅRÄZÑPÔPˆàˆˆiˆZÑàÔ$Ð0ØÔ2×9Ò9½"¼*ÑEÔEˆJˆJØÔ%Ð1Ø"&Ô":ÐØÐ"Ø&9¸!¸!¸!¸W¸*Ô&EÐ#Ø,×1Ò1°qÐ1Ñ9Ô9×@Ò@ÅÄÑLÔLˆJˆJàˆJå T°jÐAÑAÔAˆ
Ø"œmˆ
ÔØœ)ˆ
ŒØ×&Ò& yÑ1Ô1ˆ
àÐr    )NNNNN)r=   r<   r?   r>   rA   r@   rB   r@   rD   rC   rF   rH   rG   r@   r   rI   )r   rI   )r   r@   )rU   r@   r   rI   )r   r-   )r   rZ   )r^   r_   r   r   )rc   r   r   r;   )rŽ   r   r�   r�   r   r;   )r¡   r   r   r;   )r¬   r   r   r;   )r³   r   r   r;   )r»   r   r   r;   )rÄ   rÅ   r   r;   )rÞ   rß   r   rà   )rã   rä   r   r;   )rã   rô   r   rõ   )rÞ   rß   rU   rø   r   rI   )r   r;   )r   r   )r   r  r   r;   )r  r   r!  r   r"  r   r   r;   r   )r5  r6  r   r   )'Ú__name__Ú
__module__Ú__qualname__Ú__doc__Ú__annotations__r?   rA   rB   rD   r   re   rF   rN   rL   ÚpropertyrG   ÚsetterrY   r]   rb   Úclassmethodr�   r    r«   r²   rº   rÃ   rÝ   râ   ró   r÷   rû   r‚   rÿ   r  r   ÚIOUr4  r>  r}   r    r   r;   r;   G   sV  € € € € € € ð`ð `ðD  ÐÐÑØ,0€HÐ0Ð0Ð0Ñ0Ø:>ÐÐ>Ð>Ð>Ñ>Ø;?ÐÐ?Ð?Ð?Ñ?Ø,0€GÐ0Ð0Ð0Ñ0Ø$˜u°TÐ:Ñ:Ô:€DÐ:Ð:Ð:Ñ:ð
 15Ø>BØ?CØ04Ø*.ð3ð 6:ð3ð 3ð 3ð 3ð 3ð 3ðj
ð 
ð 
ð 
ð ð(ð (ð (ñ „Xð(ð Ôð)ð )ð )ñ Ôð)ðð ð ð ð(ð ð ð ð0
ð 
ð 
ð 
ð$ ð]
ð ]
ð ]
ñ „[ð]
ð~ ðA
ð A
ð A
ñ „[ðA
ðF ð%Xð %Xð %Xñ „[ð%XðN ð7
ð 7
ð 7
ñ „[ð7
ðr ð0ð 0ð 0ñ „[ð0ðd ðRð Rð Rñ „[ðRðh]
ð ]
ð ]
ð ]
ð~"ð "ð "ð "ð0s
ð s
ð s
ð s
ðj,"ð ,"ð ,"ð ,"ð\!ð !ð !ð !ðF ð=ð =ð =ñ „[ð=ð$ð ð ð ð$ ðu
ð u
ð u
ñ „[ðu
ðr Ø$Ø(5Ô(9ð	`!ð `!ð `!ð `!ð `!ðF AEðQð Qð Qð Qð Qð Qð Qr    r;   )r   r   r   r   r   r   )r!   r"   r   r#   )6Ú
__future__r   ÚloggingÚcollections.abcr   r   Údataclassesr   r   Útypingr   r	   r§   r   Únumpy.typingrÎ   Úsupervision.configr
   Úsupervision.detection.corer   Ú#supervision.detection.utils._typingr   r   Ú$supervision.detection.utils.internalr   r   r   Ú'supervision.detection.utils.iou_and_nmsr   r   Úsupervision.utils.internalr   Úsupervision.validatorsr   Ú	getLoggerr?  Úloggerr-   rÑ   r6   r   rÏ   r  ré   ÚIndex1Drê   ÚIndex2Dr.   r  r"   r#   r   r8   r;   r}   r    r   ú<module>rY     sx  ðØ "Ð "Ð "Ð "Ð "Ð "à €€€Ø .Ð .Ð .Ð .Ð .Ð .Ð .Ð .Ø (Ð (Ð (Ð (Ð (Ð (Ð (Ð (Ø Ð Ð Ð Ð Ð Ð Ð à Ð Ð Ð Ø Ð Ð Ð Ð Ð à 4Ð 4Ð 4Ð 4Ð 4Ð 4Ø 1Ð 1Ð 1Ð 1Ð 1Ð 1ð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 7Ð 6Ð 6Ð 6Ð 6Ð 6Ø =Ð =Ð =Ð =Ð =Ð =à	ˆÔ	˜8Ñ	$Ô	$€ð ˆ%�K�$�s”)Ñ˜d 4œjÑ(¨3¬;°r´wÔ+?Ñ?À#Ä+ÈbÌhÔBWÑWð ð �˜Ð Ô
!€Ø�r”z #”Ñ&¨¬°R´ZÔ)@Ñ@À4ÈÄ9ÑLÈuÑT€Ø”k "¤*Ô-°°S´	Ñ9¸EÑAÐ ð)ð )ð )ð )ðð ð ð ð. €�ÐÑÔðSð Sð Sð Sð Sñ Sô Sñ ÔðSð Sð Sr    