o hf2@sdZddlmZddlmZmZddlZddlmZddlm Z ddl m Z dd l m Z mZmZmZd d lmZd gZd ZeGddde ZGddde ZGdddejZGdddejZGdddejZGdddejZGdddejZdZdZedeGdd d eZdS)!z PyTorch ViTMatte model.) dataclass)OptionalTupleN)nn) AutoBackbone)PreTrainedModel) ModelOutputadd_start_docstrings%add_start_docstrings_to_model_forwardreplace_return_docstrings)VitMatteConfigz$hustvl/vitmatte-small-composition-1krc@s^eZdZUdZdZeejed<dZ ejed<dZ ee ejed<dZ ee ejed<dS)ImageMattingOutputa Class for outputs of image matting models. Args: loss (`torch.FloatTensor` of shape `(1,)`, *optional*, returned when `labels` is provided): Loss. alphas (`torch.FloatTensor` of shape `(batch_size, num_channels, height, width)`): Estimated alpha values. hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`): Tuple of `torch.FloatTensor` (one for the output of the embeddings, if the model has an embedding layer, + one for the output of each stage) of shape `(batch_size, sequence_length, hidden_size)`. Hidden-states (also called feature maps) of the model at the output of each stage. attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`): Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, patch_size, sequence_length)`. Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads. Nlossalphas hidden_states attentions) __name__ __module__ __qualname____doc__rrtorch FloatTensor__annotations__rrrrrrd/var/www/html/ai/venv/lib/python3.10/site-packages/transformers/models/vitmatte/modeling_vitmatte.pyr,s rc@s$eZdZdZeZdZdZddZdS)VitMattePreTrainedModelz An abstract class to handle weights initialization and a simple interface for downloading and loading pretrained models. pixel_valuesTcCsDt|tjr|jjjd|jjd|jdur |jj dSdSdS)Ng)meanstd) isinstancerConv2dweightdatanormal_configinitializer_rangebiaszero_)selfmodulerrr _init_weightsRs  z%VitMattePreTrainedModel._init_weightsN) rrrrr config_classmain_input_namesupports_gradient_checkpointingr,rrrrrHs  rcs*eZdZdZdfdd ZddZZS) VitMatteBasicConv3x3zP Basic convolution layers including: Conv3x3, BatchNorm2d, ReLU layers. r csBttj||d||dd|_tj||jd|_t|_ dS)NrF) in_channels out_channels kernel_sizestridepaddingr()eps) super__init__rr"conv BatchNorm2dbatch_norm_eps batch_normReLUrelu)r*r&r2r3r5r6 __class__rrr9^s zVitMatteBasicConv3x3.__init__cCs"||}||}||}|SN)r:r=r?r* hidden_staterrrforwardks   zVitMatteBasicConv3x3.forward)r1r rrrrr9rE __classcell__rrr@rr0Ys r0c(eZdZdZfddZddZZS)VitMatteConvStreamzc Simple ConvStream containing a series of basic conv3x3 layers to extract detail features. csvt|jj}|j}t|_|g||_t t |jdD]}|j|}|j|d}|j t |||q dS)Nr ) r8r9backbone_config num_channelsconvstream_hidden_sizesr ModuleListconvs conv_chansrangelenappendr0)r*r&r2r3iin_chan_ out_chan_r@rrr9xs    zVitMatteConvStream.__init__cCsJd|i}|}tt|jD]}|j||}dt|d}|||<q |S)Ndetailed_feature_map_0detailed_feature_map_r )rPrQrNstr)r*rout_dict embeddingsrSname_rrrrEs zVitMatteConvStream.forwardrFrrr@rrIss rIcrH)VitMatteFusionBlockz\ Simple fusion block to fuse features from ConvStream and Plain Vision Transformer. cs"tt|||ddd|_dS)Nr )r5r6)r8r9r0r:)r*r&r2r3r@rrr9s zVitMatteFusionBlock.__init__cCs4tjj|dddd}tj||gdd}||}|S)Nr1bilinearF) scale_factormode align_cornersr )dim)r functional interpolatercatr:)r*featuresdetailed_feature_mapupscaled_featuresoutrrrrEs zVitMatteFusionBlock.forwardrFrrr@rr\s r\crH) VitMatteHeadzJ Simple Matting Head, containing only conv3x3 and conv1x1 layers. c sZt|jd}d}ttj||ddddt|tdtj|ddddd|_dS)Nrr )r4r5r6Tr) r8r9fusion_hidden_sizesr Sequentialr"r;r> matting_convs)r*r&r2 mid_channelsr@rrr9s   zVitMatteHead.__init__cCs||}|SrB)rnrCrrrrEs zVitMatteHead.forwardrFrrr@rris  ricrH)VitMatteDetailCaptureModulezG Simple and lightweight Detail Capture Module for ViT Matting. c stt|jt|jdkrtd||_t||_|jj |_ t |_ |j g|j|_tt|jdD]}|j t||j||j |d |j|ddq8t||_dS)Nr z_The length of fusion_hidden_sizes should be equal to the length of convstream_hidden_sizes + 1.)r&r2r3)r8r9rQrlrL ValueErrorr&rI convstreamrOrrM fusion_blocks hidden_sizefusion_channelsrPrRr\ri matting_head)r*r&rSr@rrr9s&     z$VitMatteDetailCaptureModule.__init__cCs`||}tt|jD]}dtt|j|d}|j||||}q t||}|S)NrWr )rrrPrQrsrXrsigmoidrv)r*rerdetail_featuresrSdetailed_feature_map_namerrrrrEs z#VitMatteDetailCaptureModule.forwardrFrrr@rrps rpaI Parameters: This model is a PyTorch [torch.nn.Module](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) sub-class. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and behavior. config ([`UperNetConfig`]): Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the [`~PreTrainedModel.from_pretrained`] method to load the model weights. aw Args: pixel_values (`torch.FloatTensor` of shape `(batch_size, num_channels, height, width)`): Pixel values. Padding will be ignored by default should you provide it. Pixel values can be obtained using [`AutoImageProcessor`]. See [`VitMatteImageProcessor.__call__`] for details. output_attentions (`bool`, *optional*): Whether or not to return the attentions tensors of all attention layers in case the backbone has them. See `attentions` under returned tensors for more detail. output_hidden_states (`bool`, *optional*): Whether or not to return the hidden states of all layers of the backbone. See `hidden_states` under returned tensors for more detail. return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. zNViTMatte framework leveraging any vision backbone e.g. for ADE20k, CityScapes.csxeZdZfddZeedeee d     d de e j de e de e d e e j d e e f d d ZZS)VitMatteForImageMattingcs6t|||_t|j|_t||_| dSrB) r8r9r&r from_configrJbackbonerpdecoder post_init)r*r&r@rrr9s   z VitMatteForImageMatting.__init__zbatch_size, sequence_length) output_typer-Nroutput_attentionsoutput_hidden_stateslabels return_dictc Cs|dur|n|jj}|dur|n|jj}|dur|n|jj}|jj|||d}|jd}|||}d} |dur>> from transformers import VitMatteImageProcessor, VitMatteForImageMatting >>> import torch >>> from PIL import Image >>> from huggingface_hub import hf_hub_download >>> processor = VitMatteImageProcessor.from_pretrained("hustvl/vitmatte-small-composition-1k") >>> model = VitMatteForImageMatting.from_pretrained("hustvl/vitmatte-small-composition-1k") >>> filepath = hf_hub_download( ... repo_id="hf-internal-testing/image-matting-fixtures", filename="image.png", repo_type="dataset" ... ) >>> image = Image.open(filepath).convert("RGB") >>> filepath = hf_hub_download( ... repo_id="hf-internal-testing/image-matting-fixtures", filename="trimap.png", repo_type="dataset" ... ) >>> trimap = Image.open(filepath).convert("L") >>> # prepare image + trimap for the model >>> inputs = processor(images=image, trimaps=trimap, return_tensors="pt") >>> with torch.no_grad(): ... alphas = model(**inputs).alphas >>> print(alphas.shape) torch.Size([1, 1, 640, 960]) ```N)rrrjzTraining is not yet supportedr )rrrr) r&use_return_dictrrr|forward_with_filtered_kwargs feature_mapsr}NotImplementedErrorrrr) r*rrrrroutputsrerroutputrrrrE s*,  zVitMatteForImageMatting.forward)NNNNN)rrrr9r VITMATTE_INPUTS_DOCSTRINGformatr r_CONFIG_FOR_DOCrrTensorboolrErGrrr@rrzs(  rz) r dataclassesrtypingrrrrrmodeling_utilsrutilsr r r r configuration_vitmatter&VITMATTE_PRETRAINED_MODEL_ARCHIVE_LISTrrrModuler0rIr\rirpVITMATTE_START_DOCSTRINGrrzrrrrs6     )