hI&SrSSKJr SSKJr SSKJrJr SSKr SSK J r SSK J r Jr SS KJrJrJr /S QrS r"S S \5r"SS\5r\ "S5SSj5r\ "S5SSj5rSSjr\ "S5S S!Sjj5r\ "S5S S"Sjj5r\ "S5S#Sj5r\ "S5SSj5rS$S%SjjrS&SjrS'Sjr SSjr!S(Sjr"S)Sjr#g)*aUtilities for handling bounding box operations during image augmentation. This module provides tools for processing bounding boxes in various formats (COCO, Pascal VOC, YOLO), converting between coordinate systems, normalizing and denormalizing coordinates, filtering boxes based on visibility and size criteria, and performing transformations on boxes to match image augmentations. It forms the core functionality for all bounding box-related operations in the albumentations library. ) annotations)Sequence)AnyLiteralN)handle_empty_array)MONO_CHANNEL_DIMENSIONS$NUM_BBOXES_COLUMNS_IN_ALBUMENTATIONS) DataProcessorParams ShapeType) BboxParams BboxProcessor check_bboxes"convert_bboxes_from_albumentations convert_bboxes_to_albumentationsdenormalize_bboxes filter_bboxesnormalize_bboxesunion_of_bboxesc^\rSrSrSrS S U4SjjjrS U4Sjjr\S Sj5r\S Sj5r S Sjr Sr U=r $)r%a# Parameters for bounding box transforms. Args: format (Literal["coco", "pascal_voc", "albumentations", "yolo"]): Format of bounding boxes. Should be one of: - 'coco': [x_min, y_min, width, height], e.g. [97, 12, 150, 200]. - 'pascal_voc': [x_min, y_min, x_max, y_max], e.g. [97, 12, 247, 212]. - 'albumentations': like pascal_voc but normalized in [0, 1] range, e.g. [0.2, 0.3, 0.4, 0.5]. - 'yolo': [x_center, y_center, width, height] normalized in [0, 1] range, e.g. [0.1, 0.2, 0.3, 0.4]. label_fields (Sequence[str] | None): List of fields that are joined with boxes, e.g., ['class_labels', 'scores']. Default: None. min_area (float): Minimum area of a bounding box. All bounding boxes whose visible area in pixels is less than this value will be removed. Default: 0.0. min_visibility (float): Minimum fraction of area for a bounding box to remain this box in the result. Should be in [0.0, 1.0] range. Default: 0.0. min_width (float): Minimum width of a bounding box in pixels or normalized units. Bounding boxes with width less than this value will be removed. Default: 0.0. min_height (float): Minimum height of a bounding box in pixels or normalized units. Bounding boxes with height less than this value will be removed. Default: 0.0. check_each_transform (bool): If True, performs checks for each dual transform. Default: True. clip (bool): If True, clips bounding boxes to image boundaries before applying any transform. Default: False. filter_invalid_bboxes (bool): If True, filters out invalid bounding boxes (e.g., boxes with negative dimensions or boxes where x_max < x_min or y_max < y_min) at the beginning of the pipeline. If clip=True, filtering is applied after clipping. Default: False. max_accept_ratio (float | None): Maximum allowed aspect ratio for bounding boxes. The aspect ratio is calculated as max(width/height, height/width), so it's always >= 1. Boxes with aspect ratio greater than this value will be filtered out. For example, if max_accept_ratio=3.0, boxes with width:height or height:width ratios greater than 3:1 will be removed. Set to None to disable aspect ratio filtering. Default: None. Note: The processing order for bounding boxes is: 1. Convert to albumentations format (normalized pascal_voc) 2. Clip boxes to image boundaries (if clip=True) 3. Filter invalid boxes (if filter_invalid_bboxes=True) 4. Apply transformations 5. Filter boxes based on min_area, min_visibility, min_width, min_height 6. Convert back to the original format Examples: >>> # Create BboxParams for COCO format with class labels >>> bbox_params = BboxParams( ... format='coco', ... label_fields=['class_labels'], ... min_area=1024, ... min_visibility=0.1 ... ) >>> # Create BboxParams that clips and filters invalid boxes >>> bbox_params = BboxParams( ... format='pascal_voc', ... clip=True, ... filter_invalid_bboxes=True ... ) >>> # Create BboxParams that filters extremely elongated boxes >>> bbox_params = BboxParams( ... format='yolo', ... max_accept_ratio=5.0, # Filter boxes with aspect ratio > 5:1 ... clip=True ... ) c >[T U]X5 X0lX@lXPlX`lXplXlXlU bU S:a [S5eXl g)N?z]max_accept_ratio must be >= 1.0 when provided, as aspect ratio is calculated as max(w/h, h/w)) super__init__min_areamin_visibility min_width min_heightcheck_each_transformclipfilter_invalid_bboxes ValueErrormax_accept_ratio) selfformat label_fieldsrrr r!r"r#r$r& __class__s X/var/www/fran/franai/venv/lib/python3.13/site-packages/albumentations/core/bbox_utils.pyrBboxParams.__init__ns^ . ,"$$8! %:"  ',[TU]5nURURURUR UR URURURS.5 U$)zGet the private dictionary representation of bounding box parameters. Returns: dict[str, Any]: Dictionary containing the bounding box parameters. )rrr r!r"r#r&) rto_dict_privateupdaterrr r!r"r#r&)r'datar*s r+r/BboxParams.to_dict_privatesaw&(  MM"&"5"5!^^"oo(,(A(A $($9$9   r-cg)zCheck if the bounding box parameters are serializable. Returns: bool: Always returns True as BboxParams is serializable. Tclss r+is_serializableBboxParams.is_serializablesr-cg)zMGet the full name of the class. Returns: str: The string "BboxParams". rr4r5s r+get_class_fullnameBboxParams.get_class_fullnamesr-cSURSURSURSURSURSUR SUR SURS 3$) NzBboxParams(format=z, label_fields=z , min_area=z, min_visibility=z , min_width=z , min_height=z, check_each_transform=z, clip=))r(r)rrr r!r"r#r's r+__repr__BboxParams.__repr__s}  _T=N=N%>$?wtyykQR T r-)r"r#r$r&rr!rr ) NrArArATFFN)r(z7Literal['coco', 'pascal_voc', 'albumentations', 'yolo']r)zSequence[Any] | NonerfloatrrBr rBr!rBr"boolr#rCr$rCr& float | None)returndict[str, Any])rErCrEstr) __name__ __module__ __qualname____firstlineno____doc__rr/ classmethodr7r:r?__static_attributes__ __classcell__r*s@r+rr%sFV.2 #%)&+)-1G1+1 1  1  11#11 $1'116*  r-rc^\rSrSrSrS S U4Sjjjr\SSj5rSSjrSSjr SSSjjr SSjr SS jr SS jr S rU=r$)raProcessor for bounding box transformations. This class handles the preprocessing and postprocessing of bounding boxes during augmentation pipeline, including format conversion, validation, clipping, and filtering. Args: params (BboxParams): Parameters that control bounding box processing. See BboxParams class for details. additional_targets (dict[str, str] | None): Dictionary with additional targets to process. Keys are names of additional targets, values are their types. For example: {'bbox2': 'bboxes'} will handle 'bbox2' as another bounding box target. Default: None. Note: The processing order for bounding boxes is: 1. Convert to albumentations format (normalized pascal_voc) 2. Clip boxes to image boundaries (if params.clip=True) 3. Filter invalid boxes (if params.filter_invalid_bboxes=True) 4. Apply transformations 5. Filter boxes based on min_area, min_visibility, min_width, min_height 6. Convert back to the original format Examples: >>> import albumentations as A >>> # Process COCO format bboxes with class labels >>> params = A.BboxParams( ... format='coco', ... label_fields=['class_labels'], ... min_area=1024, ... min_visibility=0.1 ... ) >>> processor = BboxProcessor(params) >>> >>> # Process multiple bbox fields >>> params = A.BboxParams('pascal_voc') >>> processor = BboxProcessor( ... params, ... additional_targets={'bbox2': 'bboxes'} ... ) c$>[TU]X5 gN)rr)r'paramsadditional_targetsr*s r+rBboxProcessor.__init__s 4r-cg)zkReturns the default key for bounding box data in transformations. Returns: str: The string 'bboxes'. bboxesr4r>s r+default_data_nameBboxProcessor.default_data_namesr-c^URR(a<[U4SjURR55(d Sn[U5egg)acValidates the input bounding box data. Checks that: - Bounding boxes have labels (either in the bbox array or in label_fields) - All specified label_fields exist in the data Args: data (dict[str, Any]): Dict with bounding boxes and optional label fields. Raises: ValueError: If bounding boxes don't have labels or if label_fields are invalid. c3,># UH oT;v M g7frUr4).0ir1s r+ 2BboxProcessor.ensure_data_valid..s/\C[aT C[szOYour 'label_fields' are not valid - them must have same names as params in dictN)rVr)allr%)r'r1msgs ` r+ensure_data_validBboxProcessor.ensure_data_validsD ;; # #C/\4;;C[C[/\,\,\cCS/ !-] #r-c U [UUURRURRURRURR URR S9$)aFilter bounding boxes based on size and visibility criteria. Args: data (np.ndarray): Array of bounding boxes in Albumentations format. shape (ShapeType): Shape information for validation. Returns: np.ndarray: Filtered bounding boxes that meet the criteria. )rrr r!r&)rrVrrr r!r&r'r1shapes r+filterBboxProcessor.filters[   [[));;55kk++{{--![[99  r-c DUS:XaURRS:XaUnO [UURRUSS9nURR(a:URS:a*[ R"USS2SS24SS5USS2SS24'URR (a[UUSSSSS 9nURXB5 U$URX5 URRS:XaU$[XRRU5$) a#Converts bounding boxes between formats and applies preprocessing/postprocessing. Args: data (np.ndarray): Array of bounding boxes to process. shape (ShapeType): Image shape as dict with height and width keys. direction (Literal["to", "from"]): Direction of conversion: - "to": Convert from original format to albumentations format - "from": Convert from albumentations format to original format Default: "to". Returns: np.ndarray: Processed bounding boxes. Note: When direction="to": 1. Converts to albumentations format 2. Clips boxes if params.clip=True 3. Filters invalid boxes if params.filter_invalid_bboxes=True 4. Validates remaining boxes When direction="from": 1. Validates boxes 2. Converts back to original format toalbumentationsFcheck_validityrNr rrr r!) rVr(rr#sizenpr$rcheckr)r'r1ri directionconverted_datas r+check_and_convertBboxProcessor.check_and_converts>  {{!!%55!%!AKK&&#( "{{N$7$7!$;(*q"1"u0Eq!(Lq"1"u%{{00!."#$ " JJ~ -! ! 4 ;;  !1 1K1$ 8J8JERRr-c[U5 g)zCheck if bounding boxes are valid. Args: data (np.ndarray): Array of bounding boxes to validate. shape (ShapeType): Shape to check against. N)rrhs r+ruBboxProcessor.check[s  Tr-c z[R"[XRRUSS9UR S9$)aConvert bounding boxes from internal Albumentations format to the specified format. Args: data (np.ndarray): Bounding boxes in Albumentations format. shape (ShapeType): Shape information for validation. Returns: np.ndarray: Converted bounding boxes in the target format. Trodtype)rtarrayrrVr(r~rhs r+convert_from_albumentations)BboxProcessor.convert_from_albumentationses3xx .t[[5G5G_c d**  r-c URR(a9[XRRUSS9n[ X2SSSSS9n[ U5 U$[XRRUSS9$)aConvert bounding boxes from the specified format to internal Albumentations format. Args: data (np.ndarray): Bounding boxes in source format. shape (ShapeType): Shape information for validation. Returns: np.ndarray: Converted bounding boxes in Albumentations format. FrorrrT)rVr#rr(rr)r'r1ridata_nps r+convert_to_albumentations'BboxProcessor.convert_to_albumentationsusf ;;  6t[[=O=OQVglmG#GQq\]jklG  !N/kk6H6H%`deer-r4rU)rVrrWzdict[str, str] | NonerG)r1rFrENone)r1 np.ndarrayrir rEr)rm)r1rrir rvzLiteral['to', 'from']rEr)r1rrir rEr)rIrJrKrLrMrpropertyr[rerjrxrurrrOrPrQs@r+rrs}(T55"$ 4,0 ?S?S?S) ?S  ?SB ffr-rrZc[U[5(aUSSup#O USUSp2UR5R[5nUSS2SS/4==U-ss'USS2SS/4==U-ss'U$)a*Normalize array of bounding boxes. Args: bboxes (np.ndarray): Denormalized bounding boxes `[(x_min, y_min, x_max, y_max, ...)]`. shape (ShapeType | tuple[int, int]): Image shape `(height, width)`. Returns: np.ndarray: Normalized bounding boxes `[(x_min, y_min, x_max, y_max, ...)]`. Nheightwidthrr ) isinstancetuplecopyastyperB)rZrirowscols normalizeds r+rrs{%2AY d8_eGnd%%e,Jq1a&yT!q1a&yT! r-c[U[5(a USUS4O USUS4nU[R"/UQUQS/URSS- -Q[ S9-$)a,Denormalize array of bounding boxes. Args: bboxes (np.ndarray): Normalized bounding boxes `[(x_min, y_min, x_max, y_max, ...)]`. shape (ShapeType | tuple[int, int]): Image shape `(height, width)`. Returns: np.ndarray: Denormalized bounding boxes `[(x_min, y_min, x_max, y_max, ...)]`. r rrrrqr})rrrtrrirB)rZri scale_factorss r+rrst-7ue,D,DU1XuQx(5QX>[`ai[jJkM BHH[}[}[sfllSToXYFY?Z[chi iir-c[U5S:Xa#[R"/[RS9$USUSp2USS2S4USS2S4- U-nUSS2S4USS2S4- U-nXE-$) aoCalculate areas for multiple bounding boxes. This function computes the areas of bounding boxes given their normalized coordinates and the dimensions of the image they belong to. The bounding boxes are expected to be in the format [x_min, y_min, x_max, y_max] with normalized coordinates (0 to 1). Args: bboxes (np.ndarray): A numpy array of shape (N, 4+) where N is the number of bounding boxes. Each row contains [x_min, y_min, x_max, y_max] in normalized coordinates. Additional columns beyond the first 4 are ignored. shape (ShapeType): A tuple containing the height and width of the image (height, width). Returns: np.ndarray: A 1D numpy array of shape (N,) containing the areas of the bounding boxes in pixels. Returns an empty array if the input `bboxes` is empty. Note: - The function assumes that the input bounding boxes are valid (i.e., x_max > x_min and y_max > y_min). Invalid bounding boxes may result in negative areas. - The function preserves the input array and creates a copy for internal calculations. - The returned areas are in pixel units, not normalized. Example: >>> bboxes = np.array([[0.1, 0.1, 0.5, 0.5], [0.2, 0.2, 0.8, 0.8]]) >>> image_shape = (100, 100) >>> areas = calculate_bbox_areas(bboxes, image_shape) >>> print(areas) [1600. 3600.] rr}rrNrrr )lenrtrfloat32)rZrirrwidthsheightss r+calculate_bbox_areas_in_pixelsrs< 6{axx"**--(OU7^EQTlVAqD\)U 2Fad|fQTl*f4G  r-cUS;a[SUS35eUR5R[R5n[R "U5nUSS2SS24USS2SS24'US:XaUUSS2S4USS2S4'USS2S4USS2S4'USS2S4USS2S 4-USS2S 4'USS2S4USS2S 4-USS2S 4'OUS :XaU(aD[R "USS2SS24S:*USS2SS24S:-5(a[S U35eUSS2S 4S - USS2S 4S - peUSS2S4U- USS2S4'USS2S4U- USS2S4'USS2S4U-USS2S 4'USS2S4U-USS2S 4'OUSS2SS24USS2SS24'US :wa[USS2SS24U5USS2SS24'U(a [U5 U$) a2Convert bounding boxes from a specified format to the format used by albumentations: normalized coordinates of top-left and bottom-right corners of the bounding box in the form of `(x_min, y_min, x_max, y_max)` e.g. `(0.15, 0.27, 0.67, 0.5)`. Args: bboxes (np.ndarray): A numpy array of bounding boxes with shape (num_bboxes, 4+). source_format (Literal["coco", "pascal_voc", "yolo"]): Format of the input bounding boxes. shape (ShapeType): Image shape (height, width). check_validity (bool): Check if all boxes are valid boxes. Returns: np.ndarray: An array of bounding boxes in albumentations format with shape (num_bboxes, 4+). Raises: ValueError: If `source_format` is not 'coco', 'pascal_voc', or 'yolo'. ValueError: If in YOLO format, any coordinates are not in the range (0, 1]. cocoyolo pascal_voczUnknown source_format 8. Supported formats are: 'coco', 'pascal_voc' and 'yolo'Nrqrrr rrrzFIn YOLO format all coordinates must be float and in range (0, 1], got ) r%rrrtr zeros_likeanyrr)rZ source_formatrirpconverted_bboxesw_halfh_halfs r+rrs22::$]O3k l  [[] ! !"** -F}}V,$QUmQU!'1A!'1A!'1q!t !<A!'1q!t !<A & bfffQUmq&8VArrE]Q=N%OPPeflemno o1)6!Q$QRSUVRV>W"W[\!\A"5ad";>QRSUVRV>W"W[\!\A!4QT!:=PQRTUQU=V!VA!4QT!:=PQRTUQU=V!VA #6BQB r-c :USS2SS24S:USS2SS24S:*-n[R"USS2SS24S5n[R"USS2SS24S5nX-U-n[R"U5(d[R"[R"USS9)5SSnXn/SQ[R"XE)5SSnU[R"XE)5SSn[ SUSUS US 35eUSS2S 4USS2S4:USS2S 4USS2S4:-n [R"U 5(dK[R"U )5SSnXnUS US::a[ S US 35e[ SUS 35eg)zCheck if bounding boxes are valid. Args: bboxes (np.ndarray): A numpy array of bounding boxes with shape (num_bboxes, 4+). Raises: ValueError: If any bounding box is invalid. Nrqrr axis)x_miny_minx_maxy_maxz Expected z for bbox z$ to be in the range [0.0, 1.0], got .rrz.x_max is less than or equal to x_min for bbox z.y_max is less than or equal to y_min for bbox )rtisclosercwherer%) rZin_range close_to_zero close_to_one valid_range invalid_idx invalid_bbox invalid_coord invalid_value valid_orders r+rrPsq"1"u "va!e}'9:HJJva!e}a0M::fQUmQ/L*\9K 66+  hh{ ;;bcpbqqr s  !Q$<&A,.6!Q$<&A,3NOK 66+  hh |,Q/2 * ?l1o -Ml^[\]^ ^I,WXYZZ r-c USUSp2[X5n[R"USS2SS/4SX4SS2SS/4S9USS2SS/4'[R"USS2SS/4SX$SS2SS/4S9USS2SS/4'[XA5$) aClip bounding boxes to the image shape. Args: bboxes (np.ndarray): A numpy array of bounding boxes with shape (num_bboxes, 4+). shape (ShapeType): The shape of the image/volume: - For 2D: {'height': int, 'width': int} - For 3D: {'height': int, 'width': int, 'depth': int} Returns: np.ndarray: A numpy array of bounding boxes with shape (num_bboxes, 4+). rrNrr)outr r)rrtr#r)rZrirr denorm_bboxess r+ clip_bboxesrvs(OU7^E'v5M& "ww}QAY'?E]^abde`f]fOghM!aV)!ww}QAY'?F^_bcefag^gPhiM!aV) M 11r-cSn[U5S:Xa3[R"/[RS9R SS5$[ X5n[ X5n [ X5n [U SS2SS24U5n U SS2S4U SS2S4- n U SS2S4U SS2S4- n Ub&[R"XU-- XU-- 5nX:*nO[R"U[S9nX:XU- :-XU-- U:-XU- :-XU- :-U-nU Un[U5S:Xa3[R"/[RS9R SS5$U$) a"Remove bounding boxes that either lie outside of the visible area by more than min_visibility or whose area in pixels is under the threshold set by `min_area`. Also crops boxes to final image size. Args: bboxes (np.ndarray): A numpy array of bounding boxes with shape (num_bboxes, 4+). shape (ShapeType): The shape of the image/volume: - For 2D: {'height': int, 'width': int} - For 3D: {'height': int, 'width': int, 'depth': int} min_area (float): Minimum area of a bounding box in pixels. Default: 0.0. min_visibility (float): Minimum fraction of area for a bounding box to remain. Default: 0.0. min_width (float): Minimum width of a bounding box in pixels. Default: 0.0. min_height (float): Minimum height of a bounding box in pixels. Default: 0.0. max_accept_ratio (float | None): Maximum allowed aspect ratio, calculated as max(width/height, height/width). Boxes with higher ratios will be filtered out. Default: None. Returns: np.ndarray: Filtered bounding boxes. gHz>rr}rqNrrr ) rrtrrreshaperrrmaximum ones_likerC)rZrirrr r!r&epsilondenormalized_box_areasclipped_bboxesclipped_box_areasrclipped_widthsclipped_heights aspect_ratios valid_ratiosmaskfiltered_bboxess r+rrs8G 6{axx"**-55a;;;OST;T288Bbjj ) 1 1!Q 7iZiir-cUR(dgUS:XagURSS:XaUSSS$Sn[R"USS2SS24SS9up4[R"USS2SS24SS9upVXS- nXd- nXq-S-n X-S-n X9- nXJ- nXY-nXj-n[ XS- 5U:d[ Xd- 5U:ag[R "X4XV/[RS 9$) aCalculate union of bounding boxes. Boxes could be in albumentations or Pascal Voc format. Args: bboxes (np.ndarray): List of bounding boxes erosion_rate (float): How much each bounding box can be shrunk, useful for erosive cropping. Set this in range [0, 1]. 0 will not be erosive at all, 1.0 can make any bbox lose its volume. Returns: np.ndarray | None: A bounding box `(x_min, y_min, x_max, y_max)` or None if no bboxes are given or if the bounding boxes become invalid after erosion. Nr rrqgư>rrg?r})rsrirtminmaxabsrr) rZ erosion_raterrrrrrr erosion_x erosion_ys r+rrs ;;q ||A!ay!}G66&BQB-a0LE66&AaC.q1LE ME ]F$s*I%+I E E E E 5=G#s5='9G'C 88U50 CCr-cx[UR5[:XaU[RS4n[R "USS9n[R "USS9n[R "URSS4[RS9n[[X55HunupV[R "U5(a[R "U5(d/SQX4'MF[R"U5SSS /upx[R"U5SSS /upXU S-US-/X4'M U$) aoCreate bounding boxes from binary masks (fast version) Args: masks (np.ndarray): Binary masks of shape (H, W) or (N, H, W) where N is the number of masks, and H, W are the height and width of each mask. Returns: np.ndarray: An array of bounding boxes with shape (N, 4), where each row is (x_min, y_min, x_max, y_max). .rrr rrqr})rrrr) rrirrtnewaxisrzerosint32 enumeratezipr) masksrrrZr`rowcolrrrrs r+bboxes_from_masksrs 5;;22bjj#o& 66%a D 66%a D XXu{{1~q) :F"3t?3 :Cvvc{{"&&++(FI88C=+QG4LE88C=+QG4LEuqy%!)8<<>5#<<>8<<>5   uU: ;   o2A2g6 7 *%J   #&J J QU%;<=r-)rZrrizShapeType | tuple[int, int]rEr)rZrrir rEr)F) rZrr%Literal['coco', 'pascal_voc', 'yolo']rir rprCrEr) rZrrrrir rprCrEr)rZrrEr)rArArrN)rZrrir rrBrrBr rBr!rBr&rDrEr)rZrrrBrEznp.ndarray | None)rrrEr)rZrrztuple[int, int]rEr)rrrrrEr)$rM __future__rcollections.abcrtypingrrnumpyrt"albumentations.augmentations.utilsr$albumentations.core.type_definitionsrr utilsr r r __all__BBOX_WITH_LABEL_SHAPErrrrrrrrrrrrrrrr4r-r+rs@#$An33 R R jLfMLf^H,Hj j &jjj((VH ! 8 888 8 8  88vH ! 0 080 0 0  00fH"["[JH'2'2Z%)Gj Gj GjGj Gj  Gj  Gj#GjGjT)DX@2  @( (((r-