o پi@sdZddlmZmZmZmZmZmZmZddl Z ddl m Z ddl m Z m Z zddlmZddlmZd ZWn eyAd ZYnwdd lmZmZmZmZmZdd lmZmZmZdd lmZddl m!Z!m"Z"ddl#m$Z$m%Z%m&Z&m'Z'm(Z(m)Z)m*Z*gdZ+eee!eee"ee$e%e&e'e(e)e*hZ,z ddlm-Z-e,.e-Wn eyYnwdee j/fddZ0dee j/fddZ1ddZ2eeehZ3defddZ4defddZ5ddZ6d e j/d!eee7ee7ffd"d#Zd e j/d$eee7e7fee7ffd%d&ZGd'd(d(e j/Z8Gd)d*d*e j/Z9dS)+zm PyTorch FX Based Feature Extraction Helpers Using https://pytorch.org/vision/stable/feature_extraction.html )CallableDictListOptionalUnionTupleTypeN)nn)_get_feature_info_get_return_layers)create_feature_extractor)get_graph_node_namesTF) Conv2dSameScaledStdConv2dSame CondConv2d StdConv2dSameFormat)resample_abs_pos_embedresample_abs_pos_embed_nhwcmaybe_add_mask)BilinearAttnTransform) MaxPool2dSame AvgPool2dSame)BatchNormAct2dSyncBatchNormActFrozenBatchNormAct2d GroupNormAct GroupNorm1Act LayerNormActLayerNormAct2d) register_notrace_moduleis_notrace_moduleget_notrace_modulesregister_notrace_functionis_notrace_functionget_notrace_functionsr rFeatureGraphNetGraphExtractNet) InplaceAbnmodulecCt||S)zq Any module not under timm.models.layers should get this decorator if we don't want to trace through it. ) _leaf_modulesaddr*r/L/home/ubuntu/.local/lib/python3.10/site-packages/timm/models/_features_fx.pyr!> r!cC|tvSN)r,r.r/r/r0r"Fr"cCttSr3)listr,r/r/r/r0r#Jr4r#funccCr+)zF Decorator for functions which ought not to be traced through )_autowrap_functionsr-r7r/r/r0r$Vr1r$cCr2r3)r8r9r/r/r0r%^r4r%cCr5r3)r6r8r/r/r/r0r&br4r&modelreturncCst|ttttddS)N leaf_modulesautowrap_functions tracer_kwargs)_get_graph_node_namesr6r,r8)r:r/r/r0rfsr return_nodescCs(tsJdt||ttttddS)NKPlease update to PyTorch 1.10+, torchvision 0.11+ for FX feature extractionr<r?)has_fx_feature_extraction_create_feature_extractorr6r,r8)r:rBr/r/r0r ms r c sfeZdZUdZejjeed<   dde j de e dfd e ed edef fd d Zd dZZS)r'zX A FX Graph based feature extractor that works with the model feature_info metadata return_dictNNCHWFr: out_indices.out_map output_fmtcsjtts Jdt|||_|durt|t|ksJt||_t|j|}t |||_ ||_ dS)NrC) super__init__rDr feature_infolenrrJr r graph_modulerF)selfr:rHrIrJrFrB __class__r/r0rLzs       zFeatureGraphNet.__init__cCs ||}|jr |St|Sr3)rOrFr6valuesrPxoutr/r/r0forwards  zFeatureGraphNet.forward)NrGF)__name__ __module__ __qualname____doc__torchjitFinalbool__annotations__r ModulerintrrstrrLrW __classcell__r/r/rQr0r'us$  r'c s~eZdZUdZejjeed<  d de j de e e e fee fdedeffdd Zd e eejejffd d ZZS)r(a A standalone feature extraction wrapper that maps dict -> list or single tensor NOTE: * one can use feature_extractor directly if dictionary output is desired * unlike FeatureGraphNet, this is intended to be used standalone and not with model feature_info metadata for builtin feature extraction mode * create_feature_extractor can be used directly if dictionary output is acceptable Args: model: model to extract features from return_nodes: node names to return features from (dict or list) squeeze_out: if only one output, and output in list format, flatten to single tensor return_dict: return as dictionary from extractor with node names as keys, ignores squeeze_out arg rFTFr:rB squeeze_outcs&t||_t|||_||_dSr3)rKrLrer rOrF)rPr:rBrerFrQr/r0rLs   zGraphExtractNet.__init__r;cCs>||}|jr |St|}|jrt|dkr|dS|S)Nr r)rOrFr6rSrerNrTr/r/r0rWs  zGraphExtractNet.forward)TF)rXrYrZr[r\r]r^r_r`r rarrrcrrLTensorrWrdr/r/rQr0r(s  & r():r[typingrrrrrrrr\r _featuresr r %torchvision.models.feature_extractionr rErrArD ImportError timm.layersrrrrrrrrtimm.layers.non_local_attnrtimm.layers.pool2d_samerrtimm.layers.norm_actrrrrrrr __all__r,r)r-rar!r"r#r8r$r%r&rcr'r(r/r/r/r0sb$     $    $(