j ,SrSSKJr SSKrSSKJr SSKJr SSK J r SSK J r SS K JrJr SS KJrJrJrJrJrJr SS KJrJr SS KJrJrJrJr \R>"\ 5r!"S S\SS9r"\"SS\55r#S/r$g)z$Image processor class for MobileViT.)UnionN) functional)TorchvisionBackend)'SemanticSegmentationPostProcessorOutput) BatchFeature)group_images_by_shapereorder_images)IMAGENET_STANDARD_MEANIMAGENET_STANDARD_STDChannelDimension ImageInputPILImageResamplingSizeDict) ImagesKwargsUnpack) TensorTypeauto_docstringloggingrequires_backendsc.\rSrSr%Sr\\S'\\S'Srg)MobileVitImageProcessorKwargs-a do_flip_channel_order (`bool`, *optional*, defaults to `self.do_flip_channel_order`): Whether to flip the color channels from RGB to BGR or vice versa. 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. do_flip_channel_orderdo_reduce_labelsN)__name__ __module__ __qualname____firstlineno____doc__bool__annotations____static_attributes__rC:\Users\johnn\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.13_qbz5n2kfra8p0\LocalCache\local-packages\Python313\site-packages\transformers/models/mobilevit/image_processing_mobilevit.pyrr-s r%rF)totalc^\rSrSrSr\r\Rr \ r \ r SS0rSrSSS.rSrSrSrS rS rSrSrS \\4U4S jjr\S+S \S \S -S \\S\4U4Sjjj5rS+S \S \S -S\S\S\ \!-S -S\"\ S4S -S\4Sjjr#S\$SS\$S4Sjr%S,Sjr&S-S \$SS\S\'SSS\S\'S \S!\(S"\S -S#\S$\S\$S4S%jjr)S.S&\$\*\+\+4S -S'\SS(4S)jjr,S*r-U=r.$)/MobileViTImageProcessor;zSTorchvision backend for MobileViT with flip_channel_order and reduce_label support. shortest_edgeF)heightwidthTNkwargsc &>[TU]"S0UD6 g)Nr)super__init__)selfr0 __class__s r&r3 MobileViTImageProcessor.__init__Os "6"r%imagessegmentation_mapsreturnc &>[TU]"X40UD6$)zX segmentation_maps (`ImageInput`, *optional*): The segmentation maps to preprocess. )r2 preprocess)r4r7r8r0r5s r&r;"MobileViTImageProcessor.preprocessRsw!&FvFFr%do_convert_rgbinput_data_formatreturn_tensorsdevicez torch.devicec URXXFS9nUR5nSUS'0n UR"U40UD6U S'UbURUSS[RS9n UR5n U R SS[ RS.5 UR"S SU 0U D6n U V s/sH1n U RS 5R[R5PM3 n n XS '[XS 9$s sn f) z"Handle extra inputs beyond images.)r7r=r>r@Fr pixel_values)r7expected_ndimsr=r>) do_rescalerresampler7rlabels)data tensor_typer) _prepare_image_like_inputscopy _preprocessr FIRSTupdaterNEARESTsqueezetotorchint64r) r4r7r8r=r>r?r@r0 images_kwargsrHprocessed_segmentation_mapssegmentation_maps_kwargsprocessed_segmentation_maps r&_preprocess_image_like_inputs5MobileViTImageProcessor._preprocess_image_like_inputs_s/00L]1  ,1 ()#//H-H^  (*.*I*I( $"2"8"8 +J+ '(.{{} $ $ + +"'-2 2 : :  +/*:*:+2+6N+ ' 3N+2M.+221588E2M (+9NBB +s,8C3rG torch.Tensorc b[[U55HnXn[R"US:H[R"SUR UR S9U5nUS- n[R"US:H[R"SUR UR S9U5nX1U'M U$)z/Reduce label values by 1, replacing 0 with 255.r)dtyper@)rangelenrRwheretensorr]r@)r4rGidxlabels r& reduce_label$MobileViTImageProcessor.reduce_labelsV%CKEKK ELLEKKX]XdXd,eglmEAIEKK ell3ekkZ_ZfZf.ginoE3K &  r%cURS:XaUR5nU/SQUSS&U$URS:Xa&UR5nUSS2/SQ4USS2SS24'U$U$)zFlip RGB to BGR or vice versa.r)rCr^rrN)ndimclone)r4r7flippeds r&flip_channel_order*MobileViTImageProcessor.flip_channel_ordersi ;;! llnG!),GAaLN [[A llnG$Q \2GAqsFON r% do_resizesizerFz7PILImageResampling | tvF.InterpolationMode | int | Nonedo_center_crop crop_sizerErescale_factordisable_groupingrrc  U (aURU5n[XS9up0nU R5H$unnU(aURUX45nUUU'M& [ X5n[UU S9up0nU R5HUunnU(aUR UU5nU(aUR UU5nU (aURU5nUUU'MW [ UU5nU$)z#Custom preprocessing for MobileViT.)rt)rfr itemsresizer center_croprescalerm)r4r7rorprFrqrrrErsrtrrr0grouped_imagesgrouped_images_indexresized_images_groupedshapestacked_imagesresized_imagesprocessed_images_groupedprocessed_imagess r&rL#MobileViTImageProcessor._preprocesss &&v.F/DV/o,!#%3%9%9%; !E>!%^T!L,: "5 )&<((>U/D^fv/w,#% %3%9%9%; !E>!%!1!1.)!L!%nn!M$!%!8!8!H.< $U +&<**BDXYr% target_sizesreturn_segmentation_scoreszBlist[torch.Tensor] | list[SemanticSegmentationPostProcessorOutput]c [US5 URn[U5nUb[U5[U5:wa [S5e[ U[ R 5(aUR5n/n[U5Hln[ RRRXGRSS9X'SSS9nUR[USRSS9USS.S 95 Mn O6[U5V s/sH n [XIRSS9XIS.S 9PM" nn U(dUV s/sHoR PM nn U$s sn fs sn f) a1 Converts the output of [`MobileViTForSemanticSegmentation`] into semantic segmentation maps. Args: outputs ([`MobileViTForSemanticSegmentation`]): Raw outputs of the model. target_sizes (`list[tuple[int, int]]`, *optional*): List of tuples corresponding to the requested final size (height, width) of each prediction. return_segmentation_scores (`bool`, *optional*, defaults to `False`): Whether to return segmentation scores alongside the segmentation map. When `True`, each element of the returned list is a [`SemanticSegmentationPostProcessorOutput`] with fields `segmentation` (class IDs, shape `(height, width)`) and `segmentation_scores` (shape `(num_classes, height, width)`). Returns: `list[torch.Tensor]` or `list[SemanticSegmentationPostProcessorOutput]`: When `return_segmentation_scores=False` (default), a list of length `batch_size` where each item is a segmentation map of shape `(height, width)` with class IDs. When `return_segmentation_scores=True`, a list of [`SemanticSegmentationPostProcessorOutput`] with fields `segmentation` (class IDs, shape `(height, width)`) and `segmentation_scores` (shape `(num_classes, height, width)`). In both cases, `(height, width)` corresponds to the target size (if `target_sizes` is specified). rRzTMake sure that you pass in as many target sizes as the batch dimension of the logitsr)dimbilinearF)rpmode align_corners) segmentationsegmentation_scores)rH)rlogitsra ValueError isinstancerRTensortolistr`nnr interpolate unsqueezeappendrargmaxr) r4outputsrrr batch_sizesemantic_segmentationrdresized_logitsiitems r&"post_process_semantic_segmentation:MobileViTImageProcessor.post_process_semantic_segmentationsr6 $([  #6{c,// j, 55+224 $& !Z(!&!4!4!@!@K))a)0|7Hzin"A"&,,;,:1,=,D,D,D,K3A!3D )"z* %+A8*0)*:*:q*:*AZ`Zcd+ "%*CX$YCX4%6%6CX !$Y$$%%Zs 'E8Er)N)r7rZr9rZ)FT)NF)/rrrr r!r valid_kwargsrBICUBICrFr image_meanr image_stdrpdefault_to_squarerrrorqrE do_normalizer=rrrr3rrrr;r"r strrrrXlistrfrmrfloatrLtupleintrr$ __classcell__)r5s@r&r)r);s.]0L!))H'J%I S !D-IINJLN #(E!F#04 G G&, G67 G  G G&59/C/C&,/C /C , /C j(4/ /Cc>)*T1/C /Cb4#7Drs+ 7;O2E5   H % L Q%0Q%Q%h % %r%