GiZddlZddlZddlmZddlmZmZddlmZddl m Z m Z ddl Z ddl ZddlmZmZmZmZmZmZmZmZmZddlmZmZmZmZmZmZdd l m!Z!erddl"Z#ddl$Z#e#jJjLZ'erdd l(m)Z)m*Z*dd l+m,Z,dd l-m.Z.e'j^e,j`e'jbe,jbe'jde,jde'jfe,jfe'jhe,jhe'jje,jjiZ6e6joDcic]\}}|| c}}Z8niZ6iZ8erddl9Z9ejte;Zd e>ejze>dfZ?GddeZ@GddeZAeBeCeDeCze>eBzfZEdZFGddeZGdZHdZIde>fdZJdZKdZLdZMdejzdeNfdZOdXd eDde>e?fd!ZP dXde>e?e?zd eDde?fd"ZQ dXde>e?e?zd eDde>e?fd#ZRdejzfd$ZS dYdejzd%eDeTeDd&fzdzde@fd'ZUdYdejzd(e@eCzdzdeDfd)ZVdYdejzd*e@dzdeTeDeDffd+ZWd,eTeDeDfd-eDd.eDdeTeDeDffd/ZXd0ee de>e fd1ZYe@jfde>e dejzfd(eCe@zde>eDfd2Z[d3eBeCe>eTzfdeNfd4Z\d3eBeCe>eTzfdeNfd5Z]d6eeBeCe>eTzfdeNfd7Z^d6eeBeCe>eTzfdeNfd8Z_ dYde eCd fd9e`dzdd fd:Zae!d;< dYde eCd fd9e`dzddfd=Zb dYde e>eTeCd fd9e`dzde d e>d e>e>d ffd>Zc dZd?eNdzd@e`dzdAeNdzdBe`e>e`zdzdCe`e>e`zdzdDeNdzdEeBeCeDfeDzdzdFeNdzdGeBeCeDfdzdHeNdzdIeBeCeDfdzdJe dKdLeDfdzfdMZdGdNdOZedPeAdQeTeAd&fd6e>eBddfdRZfdSe>eCdTe>eCfdUZgeGdVdWZhycc}}w)[N)Iterable) dataclassfields)BytesIO)AnyUnion) ExplicitEnumis_numpy_arrayis_torch_availableis_torch_tensoris_torchvision_availableis_vision_availableloggingrequires_backendsto_numpy)IMAGENET_DEFAULT_MEANIMAGENET_DEFAULT_STDIMAGENET_STANDARD_MEANIMAGENET_STANDARD_STDOPENAI_CLIP_MEANOPENAI_CLIP_STD)requires) ImageReadMode decode_image)InterpolationMode) pil_to_tensorzPIL.Image.Imagez torch.TensorceZdZdZdZy)ChannelDimensionchannels_first channels_lastN)__name__ __module__ __qualname__FIRSTLASTZ/var/www/pod-logistic/pod-ai/venv/lib/python3.12/site-packages/transformers/image_utils.pyrrUs E Dr(rceZdZdZdZy)AnnotationFormatcoco_detection coco_panopticN)r"r#r$COCO_DETECTION COCO_PANOPTICr'r(r)r+r+Zs %N#Mr(r+cbtxr$t|tjjSN)r isinstancePILImageimgs r) is_pil_imager7bs  EZSYY__%EEr(ceZdZdZdZdZy) ImageTypepillowtorchnumpyN)r"r#r$r3TORCHNUMPYr'r(r)r9r9fs C E Er(r9ct|rtjSt|rtjSt |rtj Stdt|)NzUnrecognized image type ) r7r9r3r r=r r> ValueErrortypeimages r)get_image_typerDlsOE}}ue /U }= >>r(cLt|xst|xs t|Sr1)r7r r r5s r)is_valid_imagerFvs!   Ks 3 Ks7KKr(imagesc.|xrtd|DS)Nc32K|]}t|ywr1)rF).0rCs r) z*is_valid_list_of_images..{sDE./Dall)rGs r)is_valid_list_of_imagesrOzs  DcDVDDDr(c8t|dtr|Dcgc] }|D]}| c}}St|dtjrtj|dSt|dt j rt j|dSycc}}w)Nraxis)dim)r2listnpndarray concatenater;Tensorcat) input_listsublistitems r)concatenate_listr]~s~*Q-&$.C7C4CCC JqM2:: .~~jq11 JqM5<< 0yy++ 1DsBcrt|ttfr|D]}t|ryyt |syy)NFT)r2rTtuple valid_imagesrF)imgsr6s r)r`r`s?$u & C$  D ! r(cLt|ttfrt|dSy)NrF)r2rTr_rFr5s r) is_batchedrcs"#e}%c!f%% r(rCreturnc|jtjk(rytj|dk\xrtj|dkS)zV Checks to see whether the pixel values have already been rescaled to [0, 1]. Frr )dtyperUuint8minmaxrBs r)is_scaled_imagerjs> {{bhh 66%=A  4"&&-1"44r(expected_ndimsc (t|r|St|r|gSt|rU|j|dzk(r t |}|S|j|k(r|g}|St d|dzd|d|jdt dt |d)a Ensure that the output is a list of images. If the input is a single image, it is converted to a list of length 1. If the input is a batch of images, it is converted to a list of images. Args: images (`ImageInput`): Image of images to turn into a list of images. expected_ndims (`int`, *optional*, defaults to 3): Expected number of dimensions for a single input image. If the input image has a different number of dimensions, an error is raised. r z%Invalid image shape. Expected either z or z dimensions, but got z dimensions.z]Invalid image type. Expected either PIL.Image.Image, numpy.ndarray, or torch.Tensor, but got .)rcr7rFndimrTr@rA)rGrks r)make_list_of_imagesros& Fxf ;;.1, ,&\F [[N *XF  78J7K4P^O_`KK= .   ghlmshtguuvw r(cHt|ttfr=td|Dr+td|Dr|Dcgc] }|D]}| c}}St|ttfr[t |rPt |ds|dj |k(r|S|dj |dzk(r|Dcgc] }|D]}| c}}St|r:t |s|j |k(r|gS|j |dzk(r t|Std|cc}}wcc}}w)a Ensure that the output is a flat list of images. If the input is a single image, it is converted to a list of length 1. If the input is a nested list of images, it is converted to a flat list of images. Args: images (`Union[list[ImageInput], ImageInput]`): The input image. expected_ndims (`int`, *optional*, defaults to 3): The expected number of dimensions for a single input image. Returns: list: A list of images or a 4d array of images. c3HK|]}t|ttfywr1r2rTr_rJimages_is r)rKz+make_flat_list_of_images..K 8dE]3K "c3<K|]}t|xs| ywr1rOrss r)rKz+make_flat_list_of_images.. Yh'1A\AYrr z*Could not make a flat list of images from r2rTr_rNrOr7rnrFr@)rGrkimg_listr6s r)make_flat_list_of_imagesr}s" 6D%=) KFK K YRXY Y$*?h?s???&4-(-DV-L q "fQinn&FM !9>>^a/ /(.CH(C3CCCC Cf  6;;.#@8O ;;.1, ,<  A&J KK@ Ds D1Dc t|ttfr&td|Drtd|Dr|St|ttfr\t |rQt |ds|dj |k(r|gS|dj |dzk(r|Dcgc] }t|c}St|r.rurvc3<K|]}t|xs| ywr1rxrss r)rKz-make_nested_list_of_images..ryrzrr z]Invalid input type. Must be a single image, a list of images, or a list of batches of images.r{)rGrkrCs r)make_nested_list_of_imagesrs 6D%=) KFK K YRXY Y &4-(-DV-L q "fQinn&F8O !9>>^a/ /-34EDK4 4f  6;;.#@H:  ;;.1, ,L> ! t uu5sDct|stdt|tr9t |t j j rtj|St|S)NzInvalid image type: ) rFr@rArr2r3r4rUarrayrr5s r)to_numpy_arrayrsP # /S {;<<C!Axx} C=r( num_channels.c*||nd}t|tr|fn|}|jdk(rd\}}nB|jdk(rd\}}n-|jdk(rd\}}ntd|j|j||vrD|j||vr3t j d|jd tjS|j||vrtjS|j||vrtjStd ) a[ Infers the channel dimension format of `image`. Args: image (`np.ndarray`): The image to infer the channel dimension of. num_channels (`int` or `tuple[int, ...]`, *optional*, defaults to `(1, 3)`): The number of channels of the image. Returns: The channel dimension of the image. r r)r)rrz(Unsupported number of image dimensions: z4The channel dimension is ambiguous. Got image shape z. Assuming channels are the first dimension. Use the [input_data_format](https://huggingface.co/docs/transformers/main/internal/image_processing_utils#transformers.image_transforms.rescale.input_data_format) parameter to assign the channel dimension.z(Unable to infer channel dimension format) r2intrnr@shapeloggerwarningrr%r&)rCr first_dimlast_dims r)infer_channel_dimension_formatr s$0#;<L&0s&CL?L zzQ" 8 q" 8 q" 8CEJJ<PQQ {{9-%++h2G<2WB5;;-PJ K  %%% Y < /%%% X , .$$$ ? @@r(input_data_formatc| t|}|tjk(r|jdz S|tjk(r|jdz St d|)a Returns the channel dimension axis of the image. Args: image (`np.ndarray`): The image to get the channel dimension axis of. input_data_format (`ChannelDimension` or `str`, *optional*): The channel dimension format of the image. If `None`, will infer the channel dimension from the image. Returns: The channel dimension axis of the image. rr Unsupported data format: )rrr%rnr&r@)rCrs r)get_channel_dimension_axisrGsd :5A,222zzA~ .33 3zzA~ 01B0CD EEr( channel_dimc| t|}|tjk(r|jd|jdfS|tjk(r|jd|jdfSt d|)a Returns the (height, width) dimensions of the image. Args: image (`np.ndarray`): The image to get the dimensions of. channel_dim (`ChannelDimension`, *optional*): Which dimension the channel dimension is in. If `None`, will infer the channel dimension from the image. Returns: A tuple of the image's height and width. r)rrr%rr&r@)rCrs r)get_image_sizer]s{4U; &,,,{{2 B// (-- -{{2 B//4[MBCCr( image_size max_height max_widthcx|\}}||z }||z }t||}t||z}t||z} || fS)a Computes the output image size given the input image and the maximum allowed height and width. Keep aspect ratio. Important, even if image_height < max_height and image_width < max_width, the image will be resized to at least one of the edges be equal to max_height or max_width. For example: - input_size: (100, 200), max_height: 50, max_width: 50 -> output_size: (25, 50) - input_size: (100, 200), max_height: 200, max_width: 500 -> output_size: (200, 400) Args: image_size (`tuple[int, int]`): The image to resize. max_height (`int`): The maximum allowed height. max_width (`int`): The maximum allowed width. )rhr) rrrheightwidth height_scale width_scale min_scale new_height new_widths r)#get_image_size_for_max_height_widthrusV,MFE&Le#KL+.IVi'(JEI%&I y  r(valuescJt|Dcgc] }t|c}Scc}w)zO Return the maximum value across all indices of an iterable of values. )zipri)rvalues_is r)max_across_indicesrs +.v, 7hCM 77 7s c*|tjk(r+t|Dcgc]}|jc}\}}}||fS|tjk(r+t|Dcgc]}|jc}\}}}||fSt d|cc}wcc}w)zH Get the maximum height and width across all images in a batch. z"Invalid channel dimension format: )rr%rrr&r@)rGrr6_rrs r)get_max_height_widthrs ,222#5F6SSsyy6S#T :y  "" .33 3#5F6SSsyy6S#T Iq  ""=>O=PQRR 7T6Ss B B annotationct|trId|vrEd|vrAt|dttfr(t |ddk(st|ddtryy)Nimage_id annotationsrTFr2dictrTr_lenrs r)"is_valid_annotation_coco_detectionrs`:t$ * $ Z ' z-04- @  =) *a /:j>WXY>Z\`3a r(ct|trMd|vrId|vrEd|vrAt|dttfr(t |ddk(st|ddtryy)Nr segments_info file_namerTFrrs r)!is_valid_annotation_coco_panopticrsh:t$ * $ z ) : % z/2T5M B  ?+ , 1Z ?@[\]@^`d5e r(rc&td|DS)Nc32K|]}t|ywr1)rrJanns r)rKz3valid_coco_detection_annotations..sN31#6NrLrMrs r) valid_coco_detection_annotationsrs N+N NNr(c&td|DS)Nc32K|]}t|ywr1)rrs r)rKz2valid_coco_panoptic_annotations..sM#05MrLrMrs r)valid_coco_panoptic_annotationsrs MM MMr(timeoutc ttdgt|tr|j ds|j drIt j jttj||dj}ntjj|r t j j|}n|j dr|jdd} t!j"|j%}t j jt|}n/t|t j j s t+d t j,j/|}|j1d }|S#t&$r}t)d |d |d }~wwxYw)a3 Loads `image` to a PIL Image. Args: image (`str` or `PIL.Image.Image`): The image to convert to the PIL Image format. timeout (`float`, *optional*): The timeout value in seconds for the URL request. Returns: `PIL.Image.Image`: A PIL Image. visionhttp://https://Trfollow_redirects data:image/,r Incorrect image source. Must be a valid URL starting with `http://` or `https://`, a valid path to an image file, or a base64 encoded string. Got . Failed with NzuIncorrect format used for image. Should be an url linking to an image, a base64 string, a local path, or a PIL image.RGB)r load_imager2str startswithr3r4openrhttpxgetcontentospathisfilesplitbase64 decodebytesencode Exceptionr@ TypeErrorImageOpsexif_transposeconvert)rCrb64es r)rrst j8*-%   I &%*:*::*FIINN7599UG^b+c+k+k#lmE WW^^E "IINN5)E . C(+ ((8 ws|4 syy / D   LL ' ' .E MM% E L  ijoipp~@~AB s3A F F?)F::F?) torchvision)backendscddl}t|trK|jds|jdrdt j ||dj }|jt||j}t|tjStjj|rt|tjS|jd r|j!d d } t#j$|j'}|jt||j}t|tjSt|t,j.j.r9t,j0j3|}t5|j7dSt9d#t($r}t+d |d |d}~wwxYw)at Loads `image` directly to a `torch.Tensor` using torchvision. Args: image (`str` or `PIL.Image.Image`): The image to convert to the PIL Image format. timeout (`float`, *optional*): The timeout value in seconds for the URL request. Returns: `torch.Tensor`: A `[C, H, W]` uint8 tensor in RGB channel order. rNrrTr)rf)moderrr rrrz`Incorrect format used for image. Should be a URL, a local path, a base64 string, or a PIL image.)r;r2rrrrr frombuffer bytearrayrgrrrrrrrrrrrr@r3r4rrrrr)rCrr;rawbufrs r)load_image_as_tensorrs"%   I &%*:*::*F))E7TJRRC"%""9S>EC-*;*;< < WW^^E "M,=,=> > . C(+ ((8 #%""9S>EC-*;*;< < E399?? + ++E2U]]5122 n    ijoipp~@~AB s<#G G(G##G(c <t|ttfrjt|rDt|dttfr+|Dcgc]}|Dcgc]}t ||c}c}}S|Dcgc]}t ||c}St ||Scc}wcc}}wcc}w)aLoads images, handling different levels of nesting. Args: images: A single image, a list of images, or a list of lists of images to load. timeout: Timeout for loading images. Returns: A single image, a list of images, a list of lists of images. r)r)r2rTr_rr)rGr image_grouprCs r) load_imagesr,s&4-( v;:fQi$?eklVa[QEZw7Ql lDJK5Jug6K K&'22 RlKs B B B*BB do_rescalerescale_factor do_normalize image_mean image_stddo_padpad_sizedo_center_crop crop_size do_resizesizeresamplePILImageResamplingrc |r | td|r | td|r|| td|r | td| r| | tdyy)a Checks validity of typically used arguments in an `ImageProcessor` `preprocess` method. Raises `ValueError` if arguments incompatibility is caught. Many incompatibilities are model-specific. `do_pad` sometimes needs `size_divisor`, sometimes `size_divisibility`, and sometimes `size`. New models and processors added should follow existing arguments when possible. Nz=`rescale_factor` must be specified if `do_rescale` is `True`.zgDepending on the model, `size_divisor` or `pad_size` or `size` must be specified if `do_pad` is `True`.zP`image_mean` and `image_std` must both be specified if `do_normalize` is `True`.z<`crop_size` must be specified if `do_center_crop` is `True`.zA`size` and `resample` must be specified if `do_resize` is `True`.)r@) rrrrrrrrrrrrs r)validate_preprocess_argumentsrAs,n,XYY (" u  +y/@kll)+WXX$*x/C\]]0Dyr(ceZdZdZdZddZdZdejde e zdejfd Z dd Z d Z dd Zdd ZdZdZddZy)ImageFeatureExtractionMixinzD Mixin that contain utilities for preparing image features. ct|tjjtjfs$t |st dt|dyy)Nz Got type zU which is not supported, only `PIL.Image.Image`, `np.ndarray` and `torch.Tensor` are.)r2r3r4rUrVr r@rAselfrCs r)_ensure_format_supportedz4ImageFeatureExtractionMixin._ensure_format_supportedtsQ%#))//2::!>?X]H^DK=)&& I_?r(Nc|j|t|r|j}t|tj r|'t|j dtj}|jdk(r$|jddvr|jddd}|r|dz}|jtj}tjj|S|S)a" Converts `image` to a PIL Image. Optionally rescales it and puts the channel dimension back as the last axis if needed. Args: image (`PIL.Image.Image` or `numpy.ndarray` or `torch.Tensor`): The image to convert to the PIL Image format. rescale (`bool`, *optional*): Whether or not to apply the scaling factor (to make pixel values integers between 0 and 255). Will default to `True` if the image type is a floating type, `False` otherwise. rrrr r)rr r<r2rUrVflatfloatingrnr transposeastypergr3r4 fromarray)rrCrescales r) to_pil_imagez(ImageFeatureExtractionMixin.to_pil_image{s %%e, 5 !KKME eRZZ ($UZZ]BKK@zzQ5;;q>V#;1a0 LL*E99&&u- - r(c|j|t|tjjs|S|j dS)z Converts `PIL.Image.Image` to RGB format. Args: image (`PIL.Image.Image`): The image to convert. r)rr2r3r4rrs r) convert_rgbz'ImageFeatureExtractionMixin.convert_rgbs8 %%e,%1L}}U##r(rCscalerdc.|j|||zS)z7 Rescale a numpy image by scale amount )r)rrCrs r)rz#ImageFeatureExtractionMixin.rescales %%e,u}r(c|j|t|tjjrt j |}t |r|j}|'t|jdtjn|}|r/|j|jtjd}|r"|jdk(r|jddd}|S)a Converts `image` to a numpy array. Optionally rescales it and puts the channel dimension as the first dimension. Args: image (`PIL.Image.Image` or `np.ndarray` or `torch.Tensor`): The image to convert to a NumPy array. rescale (`bool`, *optional*): Whether or not to apply the scaling factor (to make pixel values floats between 0. and 1.). Will default to `True` if the image is a PIL Image or an array/tensor of integers, `False` otherwise. channel_first (`bool`, *optional*, defaults to `True`): Whether or not to permute the dimensions of the image to put the channel dimension first. rp?rrr )rr2r3r4rUrr r<r integerrrfloat32rnr )rrCr channel_firsts r)rz*ImageFeatureExtractionMixin.to_numpy_arrays %%e, eSYY__ -HHUOE 5 !KKME;B?*UZZ]BJJ7PW LLbjj!99EE UZZ1_OOAq!,E r(c|j|t|tjjr|St |r|j d}|St j|d}|S)z Expands 2-dimensional `image` to 3 dimensions. Args: image (`PIL.Image.Image` or `np.ndarray` or `torch.Tensor`): The image to expand. rrQ)rr2r3r4r unsqueezerU expand_dimsrs r)rz'ImageFeatureExtractionMixin.expand_dimss_ %%e, eSYY__ -L 5 !OOA&E NN5q1E r(c|j|t|tjjr|j |d}nw|rut|t j r0|j|jt jd}n+t|r |j|jd}t|t j rt|t j s.t j|j|j}t|t j st j|j|j}nt|rddl}t||js?t|t j r|j |}n|j"|}t||js?t|t j r|j |}n|j"|}|j$dk(r)|j&ddvr||ddddfz |ddddfz S||z |z S)a  Normalizes `image` with `mean` and `std`. Note that this will trigger a conversion of `image` to a NumPy array if it's a PIL Image. Args: image (`PIL.Image.Image` or `np.ndarray` or `torch.Tensor`): The image to normalize. mean (`list[float]` or `np.ndarray` or `torch.Tensor`): The mean (per channel) to use for normalization. std (`list[float]` or `np.ndarray` or `torch.Tensor`): The standard deviation (per channel) to use for normalization. rescale (`bool`, *optional*, defaults to `False`): Whether or not to rescale the image to be between 0 and 1. If a PIL image is provided, scaling will happen automatically. T)rrrNrr)rr2r3r4rrUrVrrrr floatrrfr;rX from_numpytensorrnr)rrCmeanstdrr;s r) normalizez%ImageFeatureExtractionMixin.normalizes %%e, eSYY__ -''t' eRZZ (dBJJ/xx~,,U[[9c2::.hhsm**5;;7 U # dELL1dBJJ/+5++D1D'5<<-Dc5<<0c2::.*%**3/C&%,,s+C ::?u{{1~7DD$//3q$}3EE EDLC' 'r(c||ntj}|j|t|tj j s|j |}t|tr t|}t|tst|dk(r|rt|tr||fn |d|df}n|j\}}||kr||fn||f\}} t|tr|n|d} || k(r|S| t| | z|z } } |.|| krtd|d|| |kDrt|| z| z |} } ||kr| | fn| | f}|j||S)a Resizes `image`. Enforces conversion of input to PIL.Image. Args: image (`PIL.Image.Image` or `np.ndarray` or `torch.Tensor`): The image to resize. size (`int` or `tuple[int, int]`): The size to use for resizing the image. If `size` is a sequence like (h, w), output size will be matched to this. If `size` is an int and `default_to_square` is `True`, then image will be resized to (size, size). If `size` is an int and `default_to_square` is `False`, then smaller edge of the image will be matched to this number. i.e, if height > width, then image will be rescaled to (size * height / width, size). resample (`int`, *optional*, defaults to `PILImageResampling.BILINEAR`): The filter to user for resampling. default_to_square (`bool`, *optional*, defaults to `True`): How to convert `size` when it is a single int. If set to `True`, the `size` will be converted to a square (`size`,`size`). If set to `False`, will replicate [`torchvision.transforms.Resize`](https://pytorch.org/vision/stable/transforms.html#torchvision.transforms.Resize) with support for resizing only the smallest edge and providing an optional `max_size`. max_size (`int`, *optional*, defaults to `None`): The maximum allowed for the longer edge of the resized image: if the longer edge of the image is greater than `max_size` after being resized according to `size`, then the image is resized again so that the longer edge is equal to `max_size`. As a result, `size` might be overruled, i.e the smaller edge may be shorter than `size`. Only used if `default_to_square` is `False`. Returns: image: A resized `PIL.Image.Image`. r rz max_size = zN must be strictly greater than the requested size for the smaller edge size = )r)rBILINEARrr2r3r4rrTr_rrrr@resize) rrCrrdefault_to_squaremax_sizerrshortlongrequested_new_short new_shortnew_longs r)r'z"ImageFeatureExtractionMixin.resizesy< (389K9T9T %%e,%1%%e,E dD !;D dC CIN '1$'<d|47DQRGBT % v16&ufovuo t.8s.Cda#// L&93?RUY?Y\a?a;b8 '#66()(4@@DvG (*.1(Y2F2Q.RT\8 05 8,hPYEZ||D8|44r(c|j|t|ts||f}t|st|tj rP|j dk(r|j|}|jddvr|jddn|jdd}n|jd|jdf}|d|dz dz}||dz}|d|dz dz}||dz}t|tjjr|j||||fS|jddv}|sKt|tj r|jddd}t|r|jddd}|dk\r!||dkr|dk\r||dkr |d||||fS|jddt|d|dt|d|dfz} t|tj rt j || } nt|r|j#| } | d|dz dz} | |dz} | d |dz dz} | |dz}| d| | | |f<|| z }|| z }|| z }|| z }| dtd|t%| jd|td|t%| jd |f} | S) a Crops `image` to the given size using a center crop. Note that if the image is too small to be cropped to the size given, it will be padded (so the returned result has the size asked). Args: image (`PIL.Image.Image` or `np.ndarray` or `torch.Tensor` of shape (n_channels, height, width) or (height, width, n_channels)): The image to resize. size (`int` or `tuple[int, int]`): The size to which crop the image. Returns: new_image: A center cropped `PIL.Image.Image` or `np.ndarray` or `torch.Tensor` of shape: (n_channels, height, width). rrrr N.r)rr)rr2r_r rUrVrnrrrr3r4cropr permuteri zeros_like new_zerosrh)rrCr image_shapetopbottomleftrightr new_shape new_imagetop_pad bottom_padleft_pad right_pads r) center_cropz'ImageFeatureExtractionMixin.center_cropYs %%e,$&$1a7{1~- bMKN2q8{1~- AF #wz)8I+==> w'   Qs9??2#6??QPST]TcTcdfTginPoAo o r(c|j|t|tjjr|j |}|dddddddfS)a Flips the channel order of `image` from RGB to BGR, or vice versa. Note that this will trigger a conversion of `image` to a NumPy array if it's a PIL Image. Args: image (`PIL.Image.Image` or `np.ndarray` or `torch.Tensor`): The image whose color channels to flip. If `np.ndarray` or `torch.Tensor`, the channel dimension should be first. Nr)rr2r3r4rrs r)flip_channel_orderz.ImageFeatureExtractionMixin.flip_channel_ordersI %%e, eSYY__ -''.ETrT1aZ  r(c||ntjj}|j|t |tjjs|j |}|j ||||||S)a Returns a rotated copy of `image`. This method returns a copy of `image`, rotated the given number of degrees counter clockwise around its centre. Args: image (`PIL.Image.Image` or `np.ndarray` or `torch.Tensor`): The image to rotate. If `np.ndarray` or `torch.Tensor`, will be converted to `PIL.Image.Image` before rotating. Returns: image: A rotated `PIL.Image.Image`. )rexpandcenter translate fillcolor)r3r4NEARESTrr2rrotate)rrCanglerrCrDrErFs r)rHz"ImageFeatureExtractionMixin.rotatesn (389J9J %%e,%1%%e,E|| HVFicl  r(r1)NT)F)NTN)NrNNN)r"r#r$__doc__rrrrUrVrrrrrr$r'r?rArHr'r(r)rrose< $RZZ  @(2(hA5FIV!" r(rannotation_formatsupported_annotation_formatsc||vrtdtd||tjurt |s td|tj urt |s tdyy)NzUnsupported annotation format: z must be one of zInvalid COCO detection annotations. Annotations must a dict (single image) or list of dicts (batch of images) with the following keys: `image_id` and `annotations`, with the latter being a list of annotations in the COCO format.zInvalid COCO panoptic annotations. Annotations must a dict (single image) or list of dicts (batch of images) with the following keys: `image_id`, `file_name` and `segments_info`, with the latter being a list of annotations in the COCO format.)r@formatr+r.rr/r)rKrLrs r)validate_annotationsrOs  <<:6(BRSoRpqrr,;;;/ <B  ,:::.{;M <;r(valid_processor_keyscaptured_kwargsct|jt|}|r+dj|}tj d|dyy)Nz, zUnused or unrecognized kwargs: rm)set differencejoinrr)rPrQ unused_keysunused_key_strs r)validate_kwargsrXsJo&11#6J2KLK;/88HJKr(ceZdZUdZdZedzed<dZedzed<dZedzed<dZ edzed<dZ edzed<dZ edzed<d Z dd Z d Zd Zd ZdZdZddZdefdZy)SizeDictz> Hashable dictionary to store image size information. Nrr longest_edge shortest_edgerrcPt||r t||Std|d)NKey z not found in SizeDict.)hasattrgetattrKeyErrorrkeys r) __getitem__zSizeDict.__getitem__s. 4 4% %cU"9:;;r(cNt||rt|| t||S|Sr1r_r`)rrcdefaults r)rz SizeDict.gets* 4 '$"4"@4% %r(c#~Kt|D]+}t||j}||j|f-ywr1)rr`name)rfvals r)__iter__zSizeDict.__iter__s< "A$'Cffck! "s'==ct|j|j|j|j|j |j fSr1)hashrrr[r\rr)rs r)__hash__zSizeDict.__hash__ s<T[[$**d.?.?ASASUYUdUdfjftftuvvr(c:t||xrt||duSr1rfrbs r) __contains__zSizeDict.__contains__s tS!DgdC&8&DDr(cht||std|dtj|||y)Nr^z" is not a valid field of SizeDict.)r_raobject __setattr__)rrcvalues r) __setitem__zSizeDict.__setitem__s3tS!T#&HIJ J4e,r(cttrtk(Sttr;tfdt Dtfdt Dk(St S)Nc3JK|]}t|jywr1r`ri)rJrjrs r)rKz"SizeDict.__eq__..sE1qvv.E #c3JK|]}t|jywr1ry)rJrjothers r)rKz"SizeDict.__eq__..s#O+,qvv&Orz)r2rrZr_rNotImplemented)rr|s``r)__eq__zSizeDict.__eq__se eT ":& & eX &Et EEO06t OJ r(rdct|ttzr0t|}|jt|tdi|StS)Nr')r2rrZupdater}rr|mergeds r)__or__zSizeDict.__or__!s= eTH_ -$ZF MM$u+ &%f% %r(c|t|tr't|}|jt||StSr1)r2rrr}rs r)__ror__zSizeDict.__ror__(s0 eT "%[F MM$t* %Mr(r1)rdrZ)r"r#r$rJrr__annotations__rr[r\rrrdrrlrorqrvr~rrrr'r(r)rZrZsFC$JE3:#L#*# $M3:$!Jd ! IsTz <  "wE- r(rZ)rr1) NNNNNNNNNNNN)irrcollections.abcr dataclassesrriortypingrrrr<rUutilsr r r r rrrrrutils.constantsrrrrrrutils.import_utilsr PIL.Imager3 PIL.ImageOpsr4 Resamplingrtorchvision.iorrtorchvision.transformsr!torchvision.transforms.functionalrrG NEAREST_EXACTBOXr&HAMMINGBICUBICLANCZOSpil_torch_interpolation_mappingitemstorch_pil_interpolation_mappingr; get_loggerr"rrVrT ImageInputrr+rrrAnnotationTyper7r9rDrFrOr]r`rcboolrjror}rrr_rrrrrr%rrrrrrrrrrrrOrXrZ)kvs00r)rs $)    )--:8? ""$5$C$C 1 5 5##%6%?%?""$5$=$=""$5$=$=""$5$=$= '#9X8]8]8_&`1q!t&`#&(#&(#   H %rzz>48I3JDQSQ[Q[L\^bcq^rr | $|$ c39tDz112F  ?LEDE,  52::5$5##D8x}8c8brawaw # ~rzz12 3 #HKN^H^ # #Y # 4TE\8I3J t  $sD5L7H2I d  O(4TE\@Q;R2SOX\ON$sD5L?P:Q1RNW[N !* '' (* T\**Z #$!* '' (* T\* * %* \QU3 $s$55 63AF3 d#45tDAR