qyifC "SSKrSSKJr SSKJr SSKrSSKJs Jr SSK Jr SSK J r SSKJr SSKJrJr SS KJrJrJrJr SS KJr SS KJrJr SS KJrJrJ r J!r! "S S\SS9r"SSS\#\\#-S\$S\#S\4 Sjr%\"SS\ 55r&S/r'g)N)Iterable)Union) functional)TorchvisionBackend) BatchFeature)group_images_by_shapereorder_images)ChannelDimension ImageInputPILImageResamplingSizeDict)DepthEstimatorOutput) ImagesKwargsUnpack) TensorTypeauto_docstringis_torch_availablerequires_backendscB\rSrSr%Sr\\S'\\S'\\S'\\S'Srg) CHMv2ImageProcessorKwargs&a ensure_multiple_of (`int`, *optional*, defaults to 1): If `do_resize` is `True`, the image is resized to a size that is a multiple of this value. Can be overridden by `ensure_multiple_of` in `preprocess`. keep_aspect_ratio (`bool`, *optional*, defaults to `False`): If `True`, the image is resized to the largest possible size such that the aspect ratio is preserved. Can be overridden by `keep_aspect_ratio` in `preprocess`. do_reduce_labels (`bool`, *optional*, defaults to `self.do_reduce_labels`): Whether or not to reduce all label values of segmentation maps by 1. Usually used for datasets where 0 is used for background, and background itself is not included in all classes of a dataset (e.g. ADE20k). The background label will be replaced by 255. ensure_multiple_of size_divisorkeep_aspect_ratiodo_reduce_labelsN) __name__ __module__ __qualname____firstlineno____doc__int__annotations__bool__static_attributes__rm/home/mande/repo/quber/.venv/lib/python3.13/site-packages/transformers/models/chmv2/image_processing_chmv2.pyrr&s! r'rF)total input_image torch.Tensor output_sizermultiplereturncSSjnURSSupVUupxXu- n X- n U(a#[SU - 5[SU - 5:aU n OU n U"X-US9n U"X-US9n [XS9$)Nc[X- 5U-nUb XC:a[R"X- 5U-nXB:a[R"X- 5U-nU$N)roundmathfloorceil)valr-min_valmax_valxs r(constrain_to_multiple_of>get_resize_output_image_size..constrain_to_multiple_of@sQ #. !H ,  1; 3>*X5A ; #.)H4Ar')r-heightwidth)rN)shapeabsr) r*r,rr-r: input_height input_width output_height output_width scale_height scale_width new_height new_widths r(get_resize_output_image_sizerK:s !, 1 1"# 6L"-M!/L,K q; #a,&6"7 7&L'K),*EPXYJ()BXVI : 77r'c$^\rSrSrSr\r\Rr /SQr /SQr SSS.r Sr SrS rSrSrSrSrSrS rS rSrS rS \\4U4S jjr\S7S\S\S-S \\S\4U4Sjjj5rS7S\S\S-S\S\ S\!\"-S-S\#\!S4S-S\4Sjjr$S\%SS\%S4Sjr&S\%SS\S\S\'SSS \S!\'S"\S#\(S$\S%\(\%\(-S-S&\(\%\(-S-S'\S(\)S-S)\S*\)S-S+\S-S\4$S,jr*S7S-\%\+S-4S.jjr,S8S/SS\'SSS0\S(\)S-S'\SS4U4S1jjjr-S9S/SS*\)SS4S2jjr.S7S3S4S-\"\%\+\)\)4-S-S-S\%\/\!\"44S5jjr0S6r1U=r2$):CHMv2ImageProcessoraz0PIL backend for CHMV2 with reduce_label support.)gzG?gM?gl?)g$C?g+?gM?ir>TNFgp?kwargsc &>[TU]"S0UD6 g)Nr)super__init__)selfrP __class__s r(rSCHMv2ImageProcessor.__init__ys "6"r'imagessegmentation_mapsr.c &>[TU]"X40UD6$)zX segmentation_maps (`ImageInput`, *optional*): The segmentation maps to preprocess. )rR preprocess)rTrWrXrPrUs r(rZCHMv2ImageProcessor.preprocess|sw!&FvFFr'do_convert_rgbinput_data_formatreturn_tensorsdevicez torch.devicec URXXFS9nUR5nSUS'0n UR"U40UD6U S'UbURUSS[RS9n UR5n U R SSS.5 UR"S SU 0U D6n U V s/sH1n U R S 5R[R5PM3 n n XS '[XS 9$s sn f) z"Handle extra inputs beyond images.)rWr\r]r_Fr pixel_values)rWexpected_ndimsr\r]) do_normalize do_rescalerWrlabels)data tensor_typer) _prepare_image_like_inputscopy _preprocessr FIRSTupdatesqueezetotorchint64r) rTrWrXr\r]r^r_rP images_kwargsrgprocessed_segmentation_mapssegmentation_maps_kwargsprocessed_segmentation_maps r(_preprocess_image_like_inputs1CHMv2ImageProcessor._preprocess_image_like_inputss!00L]1  ,1 ()#//H-H^  (*.*I*I( $"2"8"8 +J+ '(.{{} $ $ + +URW,X Y*.*:*:+2+6N+ '3N+2M.+221588E2M (+9NBB +s8C$rfr+c b[[U55HnXn[R"US:H[R"SUR UR S9U5nUS- n[R"US:H[R"SUR UR S9U5nX1U'M U$)z/Reduce label values by 1, replacing 0 with 255.r)dtyper_r=)rangelenrpwheretensorrzr_)rTrfidxlabels r( reduce_label CHMv2ImageProcessor.reduce_labelsV%CKEKK ELLEKKX]XdXd,eglmEAIEKK ell3ekkZ_ZfZf.ginoE3K &  r'r do_resizesizeresamplez7PILImageResampling | tvF.InterpolationMode | int | Nonedo_center_crop crop_sizererescale_factorrd image_mean image_stdrrdo_padrdisable_groupingc U(aURU5n[UUS9unn0nUR5H%unnU(aURUUUUU S9nUUU'M' [ UU5n[UUS9unn0nUR5HQunnU(aUR UU5nUR UXXU 5nU(aURUU5nUUU'MS [ UU5nU$)zCustom preprocessing for CHMV2.)r)imagerrrr)rr itemsresizer center_croprescale_and_normalize pad_image)rTrWrrrrrrrerrdrrrrrrrrPgrouped_imagesgrouped_images_indexresized_images_groupedrAstacked_imagesresized_imagesprocessed_images_groupedprocessed_imagess r(rkCHMv2ImageProcessor._preprocesss', &&v.F0EV^n/o,,!#%3%9%9%; !E>!%(%'9&7 "-"-; "5 )&<((>@TU0E^fv/w,,#% %3%9%9%; !E>!%!1!1.)!L!77 LV_N!% !M.< $U +&<**BDXYr' target_sizescL[5(d [S5eURnUb[U5[U5:wa [ S5e[ U[ R5(aUR5n/n[[U55HNn[R"X5RSS9X%SSS9nUSRSS9nURU5 MP U$URSS9n[URS5Vs/sHoUPM nnU$s snf) a Converts the output of [`CHMv2ForSemanticSegmentation`] into semantic segmentation maps. Args: outputs ([`CHMv2ForSemanticSegmentation`]): Raw outputs of the model. target_sizes (`list[Tuple]` of length `batch_size`, *optional*): List of tuples corresponding to the requested final size (height, width) of each prediction. If unset, predictions will not be resized. Returns: semantic_segmentation: `list[torch.Tensor]` of length `batch_size`, where each item is a semantic segmentation map of shape (height, width) corresponding to the target_sizes entry (if `target_sizes` is specified). Each entry of each `torch.Tensor` correspond to a semantic class id. z:PyTorch is required for post_process_semantic_segmentationzTMake sure that you pass in as many target sizes as the batch dimension of the logitsr)dimbilinearFrmode align_cornersr=)r ImportErrorlogitsr} ValueError isinstancerpTensornumpyr|F interpolate unsqueezeargmaxappendrA) rToutputsrrsemantic_segmentationrresized_logits semantic_mapis r("post_process_semantic_segmentation6CHMv2ImageProcessor.post_process_semantic_segmentations+ "##Z[ [  #6{c,// j, 55+113 $& !S[)!"K))a)0|7Hzin" .a077A7> %,,\: *%$%+MMaM$8 !GLMbMhMhijMkGl$mGl!1%=Gl !$m$$%nsD!r antialiasc>UR(aUR(d[SUR535e[ UURUR4UUS9n[ TU]XX4S9$)a Resize an image to `(size["height"], size["width"])`. Args: image (`torch.Tensor`): Image to resize. size (`SizeDict`): Dictionary in the format `{"height": int, "width": int}` specifying the size of the output image. interpolation (`InterpolationMode`, *optional*, defaults to `InterpolationMode.BILINEAR`): `InterpolationMode` filter to use when resizing the image e.g. `InterpolationMode.BICUBIC`. antialias (`bool`, *optional*, defaults to `True`): Whether to use antialiasing when resizing the image ensure_multiple_of (`int`, *optional*): If `do_resize` is `True`, the image is resized to a size that is a multiple of this value keep_aspect_ratio (`bool`, *optional*, defaults to `False`): If `True`, and `do_resize` is `True`, the image is resized to the largest possible size such that the aspect ratio is preserved. Returns: `torch.Tensor`: The resized image. zDThe size dictionary must contain the keys 'height' and 'width'. Got )r,rr-)rr)r?r@rkeysrKrRr) rTrrrrrrr,rUs r(rCHMv2ImageProcessor.resize'sh:{{$**cdhdmdmdocpqr r2 djj1/'  w~e8~YYr'cURSSup4SnU"X25upgU"XB5upXX4n [R"X5$)aL Center pad a batch of images to be a multiple of `size_divisor`. Args: image (`torch.Tensor`): Image to pad. Can be a batch of images of dimensions (N, C, H, W) or a single image of dimensions (C, H, W). size_divisor (`int`): The width and height of the image will be padded to a multiple of this number. r<NcX[R"X- 5U-nX - nUS-nX4- nXE4$)Nrb)r3r5)rrnew_sizepad_size pad_size_leftpad_size_rights r(_get_pad/CHMv2ImageProcessor.pad_image.._get_pad_s9yy!45 DHH$MM%5N 0 0r')rAtvFpad) rTrrr?r@rpad_top pad_bottompad_left pad_rightpaddings r(rCHMv2ImageProcessor.pad_imageOsP BC(  1'v<&u;i<wwu&&r'rrcx[US5 URnUb#[U5[U5:wa [S5e/nUcS/[U5-OUn[ X25HVupVUb;[ R RRUSUSSS9R5nURSU05 MX U$) aj Converts the raw output of [`DepthEstimatorOutput`] into final depth predictions and depth PIL images. Only supports PyTorch. Args: outputs ([`DepthEstimatorOutput`]): Raw outputs of the model. target_sizes (`TensorType` or `List[Tuple[int, int]]`, *optional*): Tensor of shape `(batch_size, 2)` or list of tuples (`Tuple[int, int]`) containing the target size (height, width) of each image in the batch. If left to None, predictions will not be resized. Returns: `List[Dict[str, TensorType]]`: A list of dictionaries of tensors representing the processed depth predictions. rpNz]Make sure that you pass in as many target sizes as the batch dimension of the predicted depth)NN.rTrpredicted_depth) rrr}rziprpnnrrrnr)rTrrrresultsdepth target_sizes r(post_process_depth_estimation1CHMv2ImageProcessor.post_process_depth_estimationks( $(!11  $3+?3|CT+To 8D8LvO 44R^ "%o"D E&++77/*:]a8') NN-u5 6 #Er'rr1)Tr=F)r=)3rrr r!r"r valid_kwargsr BICUBICrrrrdefault_to_squarerrrrerdrrrrrrrrSrr rrZr%r strrrrvlistrrfloatr#rktuplerrrdictrr& __classcell__)rUs@r(rMrMasB:,L!))H&J%IC (DIINJL FNL#(A!B#04 G G&, G23 G  G G&59*C*C&,*C *C , *C j(4/ *Cc>)*T1*C *CX4#7DJN'''!4c3h#884?$F' d3 ?# $ ''r'rM)(r3collections.abcrtypingrrptorch.nn.functionalrrrtorchvision.transforms.v2rimage_processing_backendsrimage_processing_baserimage_transformsr r image_utilsr r r rmodeling_outputsrprocessing_utilsrrutilsrrrrrr#r%rKrM__all__rr'r(rs* $ 7;1EUU44VV E($8$8x}$$8$8 $8  $8Np,ppf ! !r'