NgN[ |ddlZddlmZddlmZddlmZmZmZm Z m Z m Z ddl Z ddl mZddlmcmZddlmZddlmZmZddlmZmZmZmZmZmZmZmZm Z m!Z!m"Z"dd l#m$Z$dd l%m&Z&dd l'm(Z(m)Z)m*Z*dd l+m,Z,m-Z-m.Z.d e e/e/ffdZ0de j1d e e/e/fde e/e/ffdZ2de/de/d e e/e/fde e/e/e/e/ffdZ3Gddej4Z5Gddej4Z6Gddej4Z7Gddej4Z8d>dZ9e,e9d d!d"d#$e9d d!%e9d&d'%e9d(d)%e9d*d+d,d-d./e9d0d,d12d3Z:d?d4Z;d@d6ee-d@d9Z?e-d@d:Z@e-d@d;ZAe-d@d<ZBe-d@d=ZCdS)AN)deepcopy)partial)CallableDictListOptionalTupleUnion)FinalIMAGENET_DEFAULT_MEANIMAGENET_DEFAULT_STD) PatchEmbedMlpDropPathClNormMlpClassifierHead LayerScaleget_norm_layer get_act_layerinit_weight_jaxinit_weight_vit to_2tupleuse_fused_attn)build_model_with_cfg)feature_take_indices) named_applycheckpoint_seqadapt_input_conv)generate_default_cfgsregister_modelregister_model_deprecations window_sizec*|j\}}}}||||dz|d||dz|d|}|ddddddd|d|d|}|S)aT Partition into non-overlapping windows with padding if needed. Args: x (tensor): input tokens with [B, H, W, C]. window_size (int): window size. Returns: windows: windows after partition with [B * num_windows, window_size, window_size, C]. (Hp, Wp): padded height and width before partition rrshapeviewpermute contiguous)xr#BHWCwindowss U/var/www/html/ai-engine/env/lib/python3.11/site-packages/timm/models/hieradet_sam2.pywindow_partitionr6sJAq!Q q!{1~%{1~qKN7JKXYN\]^^Aii1aAq))4466;;B AP[\]P^`abbG Nr4hwcN|\}}|jd||z|dz|dzz}||||dz||dz|d|dd}|dddddd|||d}|S)aZ Window unpartition into original sequences and removing padding. Args: x (tensor): input tokens with [B * num_windows, window_size, window_size, C]. window_size (int): window size. hw (Tuple): original height and width (H, W) before padding. Returns: x: unpartitioned sequences with [B, H, W, C]. rrr)r%r&r'r(r*)r4r#r8r1r2r0r/s r5window_unpartitionr:%s DAq aQUk!n4 AFGA Q[^+Q+a.-@+a.R]^_R`bdeeA !Q1a##..0055aArBBA Hr7r1r2returnc|d||dzz |dz}|d||dzz |dz}||z||z}}||||fS)Nrr)r1r2r#pad_hpad_wHpWps r5 _calc_padrB6sb ^a+a.0 0KN BE ^a+a.0 0KN BE YE B r5% r7c eZdZUejjeed< d dededede j ffd Z dej d ej fd Z xZS) MultiScaleAttention fused_attnNdimdim_out num_headsq_poolc2t||_||_||_||z}|dz|_t |_||_tj ||dz|_ tj |||_ dS)Ngr%) super__init__rFrGrHscalerrErInnLinearqkvproj)selfrFrGrHrIhead_dim __class__s r5rLzMultiScaleAttention.__init__@s  "i'% (** 9S'A+..Igw// r7r/r;c|j\}}}}|||||zd|jd}t j|d\}}} |j||||ddddd}||dddd}|jdd\}}||||z|jd}|dd}|dd}| dd} |j rtj ||| }n>||j z}||ddz} | d} | | z}|dd|||d}||}|S)Nr%r)r&rr)rF)r+rPreshaperHtorchunbindrIr- transposerEFscaled_dot_product_attentionrMsoftmaxrQ) rRr/r0r1r2_rPqkvattns r5forwardzMultiScaleAttention.forwardSsW 1ahhqkk!!!QUAt~rBB,sA&&1a ; " !Q2&&..q!Q::A A&&q!Q22A71Q3> #w//DII DI =  $DI  Z__ '  DI&&     8C7N:g{333TVT_TaTa1:S(9---bkmmZ((   )# $ $    8C7N:g{333TVT_TaTa1:S(9---bkmmr7r/r;c H|}||}|j|jkr_||}|jC|dddd}||dddd}|j}|jdd\}}||}}|jrAt|||\}}}} tj |ddd| d|f}t||}| |}|j^|jd|jdz|jd|jdzf}|jdd\}}t|||\}}}} |jr9t||||f}|ddd|d|ddf}||||z}||||||z}|S)Nrr%rr&)rrFrGrQrr-r#r+rrBr[padr6rbrwr:r.rrrrrr) rRr/shortcutr#r1r2r@rAr>r?s r5rczMultiScaleBlock.forwards JJqMM 8t| # #yy||Hy$#++Aq!Q7799X..66q!QBB& wqs|1AB   1#,Q;#?#? BE5a!Q5!U344A K00A IIaLL = $+A.$-2BBDDTUVDW[_[hij[kDklK>!A#&DAq#,Q;#?#? BE5   -"1kB8<z%HieraDet.__init__..*s,RRR13vbqbz??Q.RRRr7rrr)cg|]}|dzS)rr=rr/s r5rz%HieraDet.__init__..,sBBBa!eBBBr7T)img_sizerrr output_fmtdynamic_img_pad)r}r~rrrc6g|]}|Sr=)itemrs r5rz%HieraDet.__init__..Hs JJJAqvvxxJJJr7)rFrGrHr{rwr#rxryr&zblocks.)num_chs reductionmodule) pool_typerrxg{Gz?)stdskipjaxhead.fc)classifier_name)2rKrLrrlenrrrrrwrange stage_ends q_pool_blocksr patch_embedrrrrN ParameterrXzeros pos_embedpos_embed_windowlinspace Sequentialblocks feature_inforkrqappenddict num_featureshead_hidden_sizerheadinit trunc_normal_rrrrfix_init_weight isinstancefcrOweightdatamul_bias)#rRrrrrrHrrrrrIrwrrrrrrrzrrrrrrxrydepthdpr cur_stagerrGr#blockinit_fnrTs# ` r5rLzHieraDet.__init__sN #J// !),, 6{{c+......&& F   RRRRaVq8Q8QRRRF7777c$/#2#"677777777BBT_SbS-ABBB7F7K  !)%!#! $    D   /(#%!#    D "3 /ek!Y&VAU&V&V&VWW " U[ItGWXYGZ\`\lmn\o-p-p q qJJ>5!I!IJJJ moo u v vAG*95K%1#$(>#>#>aaK 1u''i'122 H 455 Q ##a&*+t/A*A*At'%#   E I K  u % % %DO##!!A ! 4DMsW[WfgpWqMsMsttt&vv!!5>=D1+  !!      > % G ! !$.d ! ; ; ;  , G ! !$"7T ! B B B & )4)=)=oo?GgyAAAG  & & &  #  " " " di!8 9 9 9jWYW`>a>a 9 IL  $ ) )/ : : : IL  " ' ' 8 8 8 8 8 9 9 9 9r7r/r;cJ|jdd\}}|j}tj|j||fd}|jd|jdz}|jd|jdz}||||fz}|dddd}||zS) Nrr%bicubic)sizemoderVr)rr&)r+rr[ interpolatertiler-)rRr/hw window_embedrtile_htile_ws r5 _pos_embedzHieraDet._pos_embedswqs|1, M$.1vINNN $ (:2(>>$ (:2(>> 1 1662B C CC %%aAq11 9}r7cd}t|jD]K\}}||jjjj|dz||jjjj|dzLdS)NcZ|tjd|zdS)Nr)div_mathsqrt)param _layer_ids r5rescalez)HieraDet.fix_init_weight..rescales( JJtyy11 2 2 2 2 2r7r) enumeraterrbrQrrrfc2)rRr layer_idlayers r5rzHieraDet.fix_init_weights} 3 3 3 )55 = =OHe GEJO*/A > > > GEIM(-x!| < < < < = =r7c ddgS)Nrrr=rRs r5no_weight_decayzHieraDet.no_weight_decays/00r7Fcoarsec&tddgS)Nz'^pos_embed|pos_embed_window|patch_embed)z^blocks\.(\d+)N)stemr)r)rRrs r5 group_matcherzHieraDet.group_matchers";-.    r7enablec||_dSrd)grad_checkpointing)rRrs r5set_grad_checkpointingzHieraDet.set_grad_checkpointings"(r7c|jjSrd)rrrs r5get_classifierzHieraDet.get_classifiers y|r7 reset_othercN||_|j|||dS)N)rr)rrreset)rRrrrs r5reset_classifierzHieraDet.reset_classifiers*&  { TTTTTr7NCHWindicesnorm stop_earlyrintermediates_onlyc|r Jd|dvs Jd|rAttj|\}} fd|D}j| } n%ttj|\}} |}|}g} t js|sj} njd| dz} t| D]I\} } | |}| |vr5|dkr| dd dd n|}| |J|r| S|| fS) aE Forward features that returns intermediates. Args: x: Input image tensor indices: Take last n blocks if int, all if None, select matching indices if sequence norm: Apply norm layer to all intermediates stop_early: Stop iterating over blocks when last desired intermediate hit output_fmt: Shape of intermediate feature outputs intermediates_only: Only return intermediate features coarse: Take coarse features (stage ends) if true, otherwise all block featrures Returns: z'normalization of features not supported)r"rz(Output format must be one of NCHW, NHWC.c*g|]}j|Sr=)r)rrrRs r5rz2HieraDet.forward_intermediates..s EEE1DOA.EEEr7Nrr"rr%r&) rrrrrrrXrh is_scriptingr r-r)rRr/r#r$r%rr&r take_indices max_index intermediatesrrblkx_outs` r5forward_intermediateszHieraDet.forward_intermediatess.BBBBBB----/Y---  V&:3t;O;OQX&Y&Y #L)EEEE EEEL 2II&:3t{;K;KW&U&U #L)   Q   OOA   9 ! ! # # 1: 1[FF[)a-0F'' , ,FAsAAL  1;v1E1E !Q1---1$$U+++  ! -r7 prune_norm prune_headc$|r3tt|j|\}}|j|}n%tt|j|\}}|jd|dz|_|r|jd||S)z@ Prune layers not required for specified intermediates. Nrr)r)rrrrrr )rRr#r0r1rr*r+s r5prune_intermediate_layersz"HieraDet.prune_intermediate_layerss  V&:3t;O;OQX&Y&Y #L) 2II&:3t{;K;KW&U&U #L)k.9q=.1  7 IOOA:O 6 6 6r7c||}||}t|jD]\}}||}|Srd)rrr r)rRr/rr-s r5forward_featureszHieraDet.forward_featuressW   Q   OOA   ,,  FAsAAAr7 pre_logitscb|r|||n||}|S)N)r6)r)rRr/r6s r5 forward_headzHieraDet.forward_heads03= ODIIaJI / / /499Q<<r7cZ||}||}|Srd)r5r8rs r5rczHieraDet.forwards-  ! !! $ $   a r7)r%rrrrrrrNr%rrrrrrrNrTrrurursrtF)T)NF)NFTr"FT)rFTT)!rerfrgrrkrr rrrir rNrlrLrXrmrrrhignorerrrrrr!rr/r3r5r8rcrnros@r5rrs #$,2,2-348(.&3 !/5,2 ,0!!%*"$'0;/5KM9M9M9M9 M9  M9  M9 S/M9 S/M9!c?M9!sCx1M9M9CHoM9#s(OM9M9M9 #38_!M9$sCx%M92 %S#X3M9<"%=M9>?M9@AM9B#CM9DEM9F"GM9Hbin-IM9JRY^,KM9M9M9M9M9M9^ELU\=== Y111 Y  D T     Y))T)T)))) YUUCUhsmUaeUUUU8<#$',1 1 |1 eCcN341  1  1  1 !%1 1  tEL!5tEL7I)I#JJ K1 1 1 1 j./$# 3S >*   &%,5<$5<%,r7rrc 6|ddddddttddd |S) Nr)r%r=)r>g?rr%r@zpatch_embed.projr) urlr input_size pool_sizecrop_pct interpolationmin_input_sizemeanr first_conv classifierr )rAkwargss r5_cfgrK s7 H)}%.B(      r7zfacebook/sam2-hiera-tinyzsam2_hiera_tiny.ptr?r) hf_hub_idhf_hub_filenamerBrC)rLrMzfacebook/sam2-hiera-smallzsam2_hiera_small.ptzfacebook/sam2-hiera-base-pluszsam2_hiera_base_plus.ptzfacebook/sam2-hiera-largezsam2_hiera_large.pt)r%rN)r%rO) rP)rLrMrFrBrCr)rr)rrBrC)zsam2_hiera_tiny.r224zsam2_hiera_tiny.r896sam2_hiera_smallsam2_hiera_base_plussam2_hiera_largezhieradet_small.untrainedc&|d|}i}|D]c\}}||r||d}n2|dd}|dd}|||<d|S)Nmodelrz mlp.layers.0zmlp.fc1z mlp.layers.1zmlp.fc2)getitems startswithreplace) state_dictrUprefixoutputr`ras r5checkpoint_filter_fnr]3s44J F  ""1 <<    &"%%AA  IIni 0 0 IIni 0 0q Mr7Fvariant pretrainedc |dd}d}d|vr|ddd}tt||ft t |t |d d |S) N out_indicesr'rsam2pretrained_strictFzimage_encoder.trunk.)r[getter)ra feature_cls)pretrained_filter_fn feature_cfg)pop setdefaultrrrr]r)r^r_rJracheckpoint_prefixs r5_create_hiera_detrkBs**]A..K  -u5552  %%9BSTTT[hGGG       r7c Vtdd}tdd|it|fi|S)N)rr&rr&)r(r rrsam2_hiera_tinyr_)rorrkr_rJ model_argss r5roroTs>\YGGGJ  d d: djIcIc\bIcIc d ddr7c Vtdd}tdd|it|fi|S)Nrr& r&r rnrQr_)rQrprqs r5rQrQZs>]kJJJJ  e eJ e$zJdJd]cJdJd e eer7c Xtddd}tdd|it|fi|S)Npr&)rr)rrHrrRr_)rRrprqs r5rRrR`sAq(KKKJ  i i idS]NhNhagNhNh i iir7c \tddddd}td d|it|fi|S) Nr&)r&$r')!+rr'rr)rrHrrrrSr_)rSrprqs r5rSrSfsQ&! J  e eJ e$zJdJd]cJdJd e eer7c Ztdddd}tdd|it|fi|S) Nrtrvrgh㈵>)rrrrzhieradet_smallr_)rrprqs r5rrrsD]kWdrvwwwJ  c c* cZHbHb[aHbHb c ccr7)r)Nrr:)Drcopyr functoolsrtypingrrrrr r rXtorch.nnrNtorch.nn.functional functionalr[ torch.jitr timm.datar r timm.layersrrrrrrrrrrr_builderr _featuresr _manipulaterrr _registryr r!r"rkr6rmr:rBrlrDrqrrrK default_cfgsr]rrirkrorQrRrSrr=r7r5rs ???????????????? AAAAAAAA__________________________+*****++++++FFFFFFFFFFYYYYYYYYYY U38_        5c? PUVY[^V^P_    "   5c? uS#sTWEW?X    66666")666rZZZZZbiZZZzbiBQQQQQryQQQj%$ D,, F !D,,--!D11--$"h  !% F!!!1&& >    s8$eeee ffff jjjj ffffddddddr7