fTh42nSrSSKJr SSKJrJrJr SSKrSSKrSSKJ r SSK J r SSK J r JrJr SS KJr S S KJr \R("\5r\"S S \ 55r\"SS\ 55rSSjr"SS\ R45r"SS\ R45r\"SS9"SS\55rSS/rg)zPyTorch VitPose model.) dataclass)OptionalTupleUnionN)nn)PreTrainedModel) ModelOutputauto_docstringlogging) load_backbone) VitPoseConfigc\rSrSr%SrSr\\R\ S'Sr \\R\ S'Sr \\ \RS4\ S'Sr \\ \RS4\ S'S rg) VitPoseEstimatorOutput'a Class for outputs of pose estimation models. Args: loss (`torch.FloatTensor` of shape `(1,)`, *optional*, returned when `labels` is provided): Loss is not supported at this moment. See https://github.com/ViTAE-Transformer/ViTPose/tree/main/mmpose/models/losses for further detail. heatmaps (`torch.FloatTensor` of shape `(batch_size, num_keypoints, height, width)`): Heatmaps as predicted by the model. hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`): Tuple of `torch.FloatTensor` (one for the output of the embeddings, if the model has an embedding layer, + one for the output of each stage) of shape `(batch_size, sequence_length, hidden_size)`. Hidden-states (also called feature maps) of the model at the output of each stage. attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`): Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, patch_size, sequence_length)`. Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads. Nlossheatmaps. hidden_states attentions)__name__ __module__ __qualname____firstlineno____doc__rrtorch FloatTensor__annotations__rrrr__static_attributes__rd/var/www/auris/envauris/lib/python3.13/site-packages/transformers/models/vitpose/modeling_vitpose.pyrr'sq()-D(5$$ %,,0Hhu(()0=AM8E%"3"3S"89:A:>Ju00#567>r!rc~\rSrSr\rSrSrSrS\ \ R\ R\ R4SS4SjrS rg) VitPosePreTrainedModelCvit pixel_valuesTmodulereturnNc[U[R[R45(a[RR UR RR[R5SURRS9RUR R5UR lURb%URRR5 gg[U[R 5(aJURRR5 UR RR#S5 gg)zInitialize the weightsg)meanstdNg?) isinstancerLinearConv2dinit trunc_normal_weightdatatorfloat32configinitializer_rangedtypebiaszero_ LayerNormfill_)selfr(s r" _init_weights$VitPosePreTrainedModel._init_weightsJs fryy"))4 5 5"$!6!6 ""%%emm43DKKDaDa"7"b$$% MM {{&   &&('  - - KK   " " $ MM   $ $S ).r!r)rrrrr config_classbase_model_prefixmain_input_namesupports_gradient_checkpointingrrr.r/r;r>r rr!r"r$r$CsD L$O&*# *E"))RYY *L$M *RV *r!r$cUS;a [S5eURS:wa [S5eURup4pVSnUS:XaSnUSS2SSS2S 4*USS2SSS2S 4'URUS XuU5nUR 5nUR 5H)upUSS2U S 4USS2U S 4'USS2U S 4USS2U S 4'M+ URX4XV45nUR S 5nU$) aFlip the flipped heatmaps back to the original form. Args: output_flipped (`torch.tensor` of shape `(batch_size, num_keypoints, height, width)`): The output heatmaps obtained from the flipped images. flip_pairs (`torch.Tensor` of shape `(num_keypoints, 2)`): Pairs of keypoints which are mirrored (for example, left ear -- right ear). target_type (`str`, *optional*, defaults to `"gaussian-heatmap"`): Target type to use. Can be gaussian-heatmap or combined-target. gaussian-heatmap: Classification target with gaussian distribution. combined-target: The combination of classification target (response map) and regression target (offset map). Paper ref: Huang et al. The Devil is in the Details: Delving into Unbiased Data Processing for Human Pose Estimation (CVPR 2020). Returns: torch.Tensor: heatmaps that flipped back to the original image )gaussian-heatmapcombined-targetz9target_type should be gaussian-heatmap or combined-targetzCoutput_flipped should be [batch_size, num_keypoints, height, width]rrFrN.) ValueErrorndimshapereshapeclonetolistflip) output_flipped flip_pairs target_type batch_size num_keypointsheightwidthchannelsoutput_flipped_backleftrights r" flip_backr[Ys$"AATUUa^__/=/C/C,JvH''(6q!$Q$|(D'Dq!$Q$|$#++JHeTN(..0"((* ,:1eS=,IAtSL)-;AtSL-IAucM*+.55zRX6`a-2226 r!c^\rSrSrSrS U4SjjrS S\RS\\RS\R4Sjjr Sr U=r $) VitPoseSimpleDecoderz Simple decoding head consisting of a ReLU activation, 4x upsampling and a 3x3 convolution, turning the feature maps into heatmaps. r)c>[TU]5 [R"5Ul[R "UR SSS9Ul[R"URRURSSSS9Ul g)NbilinearF) scale_factormode align_cornersrr kernel_sizestridepadding) super__init__rReLU activationUpsamplera upsamplingr/backbone_config hidden_size num_labelsconvr=r6 __class__s r"riVitPoseSimpleDecoder.__init__se '')++63F3FZglmII  " " . .0A0AqYZde  r! hidden_staterQcURU5nURU5nURU5nUb [X25nU$N)rkrmrqr[r=rurQrs r"forwardVitPoseSimpleDecoder.forwardsA|4 |4 99\*  ! 6Hr!)rkrqrm)r)Nrw) rrrrrrirTensorrryr __classcell__rss@r"r]r]s@   ELL hu||>T `e`l`l  r!r]cx^\rSrSrSrS\4U4SjjrS S\RS\ \R4Sjjr Sr U=r $) VitPoseClassicDecoderz Classic decoding head consisting of a 2 deconvolutional blocks, followed by a 1x1 convolution layer, turning the feature maps into heatmaps. r6c >[TU]5 [R"URR SSSSSS9Ul[R"S5Ul[R"5Ul [R"SSSSSSS9Ul [R"S5Ul [R"5Ul [R"SURSSSS9Ulg) NrGrF)rerfrgr9rrd)rhrirConvTranspose2drnrodeconv1 BatchNorm2d batchnorm1rjrelu1deconv2 batchnorm2relu2r/rprqrrs r"riVitPoseClassicDecoder.__init__s ))  " " . .1VW^c  ..-WWY ))#s!UV]bc ..-WWY IIc6#4#4!AWXY r!rurQcURU5nURU5nURU5nURU5nUR U5nUR U5nUR U5nUb [X25nU$rw)rrrrrrrqr[rxs r"ryVitPoseClassicDecoder.forwardsy||L1 |4 zz,/ ||L1 |4 zz,/ 99\*  ! 6Hr!)rrrqrrrrrw) rrrrrrrirr{rryr r|r}s@r"rrs; Z} ZELLhu||>Tr!rz? The VitPose model with a pose estimation head on top. ) custom_introc^\rSrSrS\SS4U4Sjjr\SS\RS\ \RS\ \RS \ \RS \ \ S \ \ S \ \ S\ \ \ 44S jj5rSrU=r$)VitPoseForPoseEstimationr6r)Nc>[TU]U5 [U5Ul[ URR S5(d [ S5e[ URR S5(d [ S5e[ URR S5(d [ S5eUR(a [U5O [U5Ul UR5 g)Nroz0The backbone should have a hidden_size attribute image_sizez0The backbone should have an image_size attribute patch_sizez/The backbone should have a patch_size attribute) rhrir backbonehasattrr6rIuse_simple_decoderr]rhead post_initrrs r"ri!VitPoseForPoseEstimation.__init__s  %f- t}}++];;OP Pt}}++\::OP Pt}}++\::NO O4:4M4M(0ShioSp  r!r' dataset_indexrQlabelsoutput_attentionsoutput_hidden_states return_dictcUbUOURRnUbUOURRnUbUOURRnSnUb [ S5eUR R UUUUUS9n U(aU RSOU SSn U RSn URRRSURRRS-n URRRSURRRS-n U RSSS5RU SX5R5n URXS9nU(d%U(a U4U SS-nO U4U SS-nUbU4U-$U$[!UUU R"U R$S 9$) a dataset_index (`torch.Tensor` of shape `(batch_size,)`): Index to use in the Mixture-of-Experts (MoE) blocks of the backbone. This corresponds to the dataset index used during training, e.g. For the single dataset index 0 refers to the corresponding dataset. For the multiple datasets index 0 refers to dataset A (e.g. MPII) and index 1 refers to dataset B (e.g. CrowdPose). flip_pairs (`torch.tensor`, *optional*): Whether to mirror pairs of keypoints (for example, left ear -- right ear). Examples: ```python >>> from transformers import AutoImageProcessor, VitPoseForPoseEstimation >>> import torch >>> from PIL import Image >>> import requests >>> processor = AutoImageProcessor.from_pretrained("usyd-community/vitpose-base-simple") >>> model = VitPoseForPoseEstimation.from_pretrained("usyd-community/vitpose-base-simple") >>> url = "http://images.cocodataset.org/val2017/000000039769.jpg" >>> image = Image.open(requests.get(url, stream=True).raw) >>> boxes = [[[412.8, 157.61, 53.05, 138.01], [384.43, 172.21, 15.12, 35.74]]] >>> inputs = processor(image, boxes=boxes, return_tensors="pt") >>> with torch.no_grad(): ... outputs = model(**inputs) >>> heatmaps = outputs.heatmaps ```NzTraining is not yet supported)rrrrrHrrr)rQ)rrrr)r6use_return_dictrrNotImplementedErrorrforward_with_filtered_kwargs feature_mapsrKrnrrpermuterL contiguousrrrr)r=r'rrQrrrrroutputssequence_outputrS patch_height patch_widthroutputs r"ry VitPoseForPoseEstimation.forwardsP&1% P 'tnP d^P  u,, -P P r!r)rE)r dataclassesrtypingrrrrtorch.utils.checkpointrmodeling_utilsr utilsr r r utils.backbone_utilsr configuration_vitposer get_loggerrloggerrr$r[Moduler]rr__all__rr!r"rs!)) - 20   H %  ?[? ?6*_***%P2996#BII#L d 5d  d N $%? @r!