
    Z jPf              	          S r SSKrSSKrSSKJr  SSKJr  SSKJrJr  SSK	J
r
  SSKJrJr  S	S
KJr  \R                   " \5      rS5S\R&                  S\S\S\R&                  4S jjr " S S\R.                  5      r " S S\R.                  5      r " S S\R.                  5      r " S S\R.                  5      r " S S\R.                  5      r " S S\R.                  5      r " S S\R.                  5      r " S S\R.                  5      r " S  S!\R.                  5      r \ " S" S#\
5      5       r!\ " S$ S%\!5      5       r" " S& S'\R.                  5      r# " S( S)\R.                  5      r$ " S* S+\R.                  5      r% " S, S-\R.                  5      r& " S. S/\R.                  5      r'\" S0S19 " S2 S3\!5      5       r(/ S4Qr)g)6zPyTorch GLPN model.    N)nn   )ACT2FN)BaseModelOutputDepthEstimatorOutput)PreTrainedModel)auto_docstringlogging   )
GLPNConfiginput	drop_probtrainingreturnc                    US:X  d  U(       d  U $ SU-
  nU R                   S   4SU R                  S-
  -  -   nU[        R                  " X@R                  U R
                  S9-   nUR                  5         U R                  U5      U-  nU$ )z[
Drop paths (Stochastic Depth) per sample (when applied in main path of residual blocks).

        r   r   )r   )dtypedevice)shapendimtorchrandr   r   floor_div)r   r   r   	keep_probr   random_tensoroutputs          w/root/GenerationalWealth/GenerationalWealth/venv/lib/python3.13/site-packages/transformers/models/glpn/modeling_glpn.py	drop_pathr       s    
 CxII[[^

Q 77E

5ELL YYMYYy!M1FM    c                      ^  \ rS rSrSrSS\S-  SS4U 4S jjjrS\R                  S\R                  4S jr	S\
4S	 jrS
rU =r$ )GLPNDropPath0   zXDrop paths (Stochastic Depth) per sample (when applied in main path of residual blocks).Nr   r   c                 .   > [         TU ]  5         Xl        g N)super__init__r   )selfr   	__class__s     r   r'   GLPNDropPath.__init__3   s    "r    hidden_statesc                 B    [        XR                  U R                  5      $ r%   )r   r   r   )r(   r+   s     r   forwardGLPNDropPath.forward7   s    FFr    c                      SU R                    3$ )Nzp=r   )r(   s    r   
extra_reprGLPNDropPath.extra_repr:   s    DNN#$$r    r0   r%   )__name__
__module____qualname____firstlineno____doc__floatr'   r   Tensorr-   strr1   __static_attributes____classcell__r)   s   @r   r"   r"   0   sQ    b#%$, #$ # #GU\\ Gell G%C % %r    r"   c                   2   ^  \ rS rSrSrU 4S jrS rSrU =r$ )GLPNOverlapPatchEmbeddings?   z+Construct the overlapping patch embeddings.c                    > [         TU ]  5         [        R                  " UUUUUS-  S9U l        [        R
                  " U5      U l        g )N   kernel_sizestridepadding)r&   r'   r   Conv2dproj	LayerNorm
layer_norm)r(   
patch_sizerE   num_channelshidden_sizer)   s        r   r'   #GLPNOverlapPatchEmbeddings.__init__B   sC    II"!O
	 ,,{3r    c                     U R                  U5      nUR                  u    p4nUR                  S5      R                  SS5      nU R	                  U5      nX$U4$ )NrB   r   )rH   r   flatten	transposerJ   )r(   pixel_values
embeddings_heightwidths         r   r-   "GLPNOverlapPatchEmbeddings.forwardN   sZ    YY|,
(..1e  ''*44Q:
__Z0
5((r    )rJ   rH   	r3   r4   r5   r6   r7   r'   r-   r;   r<   r=   s   @r   r?   r?   ?   s    5
4) )r    r?   c                   8   ^  \ rS rSrSrU 4S jr SS jrSrU =r$ )GLPNEfficientSelfAttentionY   zSegFormer's efficient self-attention mechanism. Employs the sequence reduction process introduced in the [PvT
paper](https://huggingface.co/papers/2102.12122).c                 8  > [         TU ]  5         X l        X0l        U R                  U R                  -  S:w  a&  [	        SU R                   SU R                   S35      e[        U R                  U R                  -  5      U l        U R                  U R                  -  U l        [        R                  " U R                  U R                  5      U l
        [        R                  " U R                  U R                  5      U l        [        R                  " U R                  U R                  5      U l        [        R                  " UR                  5      U l        X@l        US:  a6  [        R"                  " X"XDS9U l        [        R&                  " U5      U l        g g )Nr   zThe hidden size (z6) is not a multiple of the number of attention heads ()r   )rD   rE   )r&   r'   rM   num_attention_heads
ValueErrorintattention_head_sizeall_head_sizer   LinearquerykeyvalueDropoutattention_probs_dropout_probdropoutsr_ratiorG   srrI   rJ   r(   configrM   r^   sequence_reduction_ratior)   s        r   r'   #GLPNEfficientSelfAttention.__init__]   sI   &#6 d666!;#D$4$4#5 622316 
 $'t'7'7$:R:R'R#S !558P8PPYYt//1C1CD
99T--t/A/ABYYt//1C1CD
zz&"E"EF0#a'ii6NDG !ll;7DO	 (r    c                    UR                   S S n/ UQSPU R                  P7nU R                  U5      R                  U5      R	                  SS5      nU R
                  S:  aw  UR                   u  pn
UR                  SSS5      R                  XX#5      nU R                  U5      nUR                  XS5      R                  SSS5      nU R                  U5      n/ UR                   S S QSPU R                  P7nU R                  U5      R                  U5      R	                  SS5      nU R                  U5      R                  U5      R	                  SS5      n[        R                  " X|R	                  SS5      5      nU[        R                  " U R                  5      -  n[         R"                  R%                  USS9nU R'                  U5      n[        R                  " X5      nUR                  SSSS5      R)                  5       nUR+                  5       S S U R,                  4-   nUR                  U5      nU(       a  UU4nU$ U4nU$ )Nr   rB   r   dimr   )r   ra   rd   viewrQ   rj   permutereshaperk   rJ   re   rf   r   matmulmathsqrtr   
functionalsoftmaxri   
contiguoussizerb   )r(   r+   rU   rV   output_attentionsinput_shapehidden_shapequery_layer
batch_sizeseq_lenrL   kv_shape	key_layervalue_layerattention_scoresattention_probscontext_layernew_context_layer_shapeoutputss                      r   r-   "GLPNEfficientSelfAttention.forwardx   s0    $))#2.CCbC$*B*BCjj/44\BLLQPQR==10=0C0C-J)11!Q:BB:]ckM GGM2M)11*BOWWXY[\^_`M OOM:ML](("-LrL43K3KLHH]+00:DDQJ	jj/44X>HHAN !<<5H5HR5PQ+dii8P8P.QQ --//0@b/I ,,7_B%--aAq9DDF"/"4"4"6s";t?Q?Q>S"S%**+BC6G=/2 O\M]r    )rb   ra   ri   rM   re   rJ   r^   rd   rk   rj   rf   FrX   r=   s   @r   rZ   rZ   Y   s    98@  - -r    rZ   c                   .   ^  \ rS rSrU 4S jrS rSrU =r$ )GLPNSelfOutput   c                    > [         TU ]  5         [        R                  " X"5      U l        [        R
                  " UR                  5      U l        g r%   )r&   r'   r   rc   denserg   hidden_dropout_probri   )r(   rm   rM   r)   s      r   r'   GLPNSelfOutput.__init__   s4    YY{8
zz&"<"<=r    c                 J    U R                  U5      nU R                  U5      nU$ r%   r   ri   )r(   r+   input_tensors      r   r-   GLPNSelfOutput.forward   s$    

=1]3r    r   r3   r4   r5   r6   r'   r-   r;   r<   r=   s   @r   r   r      s    >
 r    r   c                   2   ^  \ rS rSrU 4S jrSS jrSrU =r$ )GLPNAttention   c                 `   > [         TU ]  5         [        UUUUS9U l        [	        XS9U l        g )N)rm   rM   r^   rn   )rM   )r&   r'   rZ   r(   r   r   rl   s        r   r'   GLPNAttention.__init__   s4    .# 3%=	
	 %VEr    c                 f    U R                  XX45      nU R                  US   U5      nU4USS  -   nU$ )Nr   r   )r(   r   )r(   r+   rU   rV   r   self_outputsattention_outputr   s           r   r-   GLPNAttention.forward   s@    yyQ;;|AF#%QR(88r    )r   r(   r   r   r=   s   @r   r   r      s    F r    r   c                   2   ^  \ rS rSrSU 4S jjrS rSrU =r$ )
GLPNDWConv   c           
      ^   > [         TU ]  5         [        R                  " XSSSSUS9U l        g )Nr   r   T)biasgroups)r&   r'   r   rG   dwconv)r(   rt   r)   s     r   r'   GLPNDWConv.__init__   s(    ii!QSIr    c                     UR                   u  pEnUR                  SS5      R                  XFX#5      nU R                  U5      nUR	                  S5      R                  SS5      nU$ )Nr   rB   )r   rQ   ru   r   rP   )r(   r+   rU   rV   r   r   rL   s          r   r-   GLPNDWConv.forward   sc    ,9,?,?)
\%//15:::U[cM2%--a0::1a@r    )r   )i   r   r=   s   @r   r   r      s    J r    r   c                   2   ^  \ rS rSrSU 4S jjrS rSrU =r$ )
GLPNMixFFN   c                   > [         TU ]  5         U=(       d    Un[        R                  " X#5      U l        [        U5      U l        [        UR                  [        5      (       a  [        UR                     U l        OUR                  U l        [        R                  " X45      U l        [        R                  " UR                  5      U l        g r%   )r&   r'   r   rc   dense1r   r   
isinstance
hidden_actr:   r   intermediate_act_fndense2rg   r   ri   )r(   rm   in_featureshidden_featuresout_featuresr)   s        r   r'   GLPNMixFFN.__init__   s    #2{ii= 1f''--'-f.?.?'@D$'-'8'8D$ii>zz&"<"<=r    c                     U R                  U5      nU R                  XU5      nU R                  U5      nU R                  U5      nU R	                  U5      nU R                  U5      nU$ r%   )r   r   r   ri   r   )r(   r+   rU   rV   s       r   r-   GLPNMixFFN.forward   s`    M2M5A00?]3M2]3r    )r   r   ri   r   r   )NNr   r=   s   @r   r   r      s    
> r    r   c                   6   ^  \ rS rSrSrU 4S jrSS jrSrU =r$ )	GLPNLayer   zCThis corresponds to the Block class in the original implementation.c                 >  > [         TU ]  5         [        R                  " U5      U l        [        UUUUS9U l        US:  a  [        U5      O[        R                  " 5       U l	        [        R                  " U5      U l
        [        X&-  5      n[        XUS9U l        g )N)rM   r^   rn   r   )r   r   )r&   r'   r   rI   layer_norm_1r   	attentionr"   Identityr   layer_norm_2r`   r   mlp)	r(   rm   rM   r^   r   rn   	mlp_ratiomlp_hidden_sizer)   s	           r   r'   GLPNLayer.__init__   s    LL5&# 3%=	
 5>Oi0LL5k56f_r    c                     U R                  U R                  U5      UUUS9nUS   nUSS  nU R                  U5      nXa-   nU R                  U R	                  U5      X#5      nU R                  U5      nX-   n	U	4U-   nU$ )N)r   r   r   )r   r   r   r   r   )
r(   r+   rU   rV   r   self_attention_outputsr   r   
mlp_outputlayer_outputs
             r   r-   GLPNLayer.forward  s    !%m,/	 "0 "
 2!4(,  >>*:;(8XXd//>N
 ^^J/
!1/G+r    )r   r   r   r   r   r   rX   r=   s   @r   r   r      s    M` r    r   c                   8   ^  \ rS rSrU 4S jr   SS jrSrU =r$ )GLPNEncoderi  c                   > [         T
U ]  5         Xl        [        R                  " SUR
                  [        UR                  5      SS9 Vs/ s H  o"R                  5       PM     nn/ n[        UR                  5       Hg  nUR                  [        UR                  U   UR                  U   US:X  a  UR                  OUR                   US-
     UR                   U   S95        Mi     ["        R$                  " U5      U l        / nSn[        UR                  5       H  n/ nUS:w  a  XqR                  US-
     -  n[        UR                  U   5       HY  n	UR                  [)        UUR                   U   UR*                  U   X7U	-      UR,                  U   UR.                  U   S95        M[     UR                  ["        R$                  " U5      5        M     ["        R$                  " U5      U l        ["        R$                  " [        UR                  5       Vs/ s H&  n["        R2                  " UR                   U   5      PM(     sn5      U l        g s  snf s  snf )Nr   cpu)r   r   )rK   rE   rL   rM   )rM   r^   r   rn   r   )r&   r'   rm   r   linspacedrop_path_ratesumdepthsitemrangenum_encoder_blocksappendr?   patch_sizesstridesrL   hidden_sizesr   
ModuleListpatch_embeddingsr   r^   	sr_ratios
mlp_ratiosblockrI   rJ   )r(   rm   xdprrS   iblockscurlayersjr)   s             r   r'   GLPNEncoder.__init__  s    "'63H3H#fmmJ\ej!kl!kAvvx!kl 
v001A*%11!4!>>!,89Q!4!4FDWDWXY\]X]D^ & 3 3A 6	 2 !#j 9 v001AFAv}}QU++6==+,$*$7$7$:,2,F,Fq,I"%Ag,171A1A!1D"("3"3A"6	 - MM"--/0! 2$ ]]6*
 --;@AZAZ;[\;[aR\\&--a01;[\
O mP ]s   I-Ic                 2   U(       a  SOS nU(       a  SOS nUR                   S   nUn[        [        U R                  U R                  U R
                  5      5       H  u  pU
u  pnU" U5      u  pn[        U5       H&  u  nnU" XX5      nUS   nU(       d  M  UUS   4-   nM(     U" U5      nUR                  X~US5      R                  SSSS5      R                  5       nU(       d  M  XX4-   nM     U(       d  [        S XU4 5       5      $ [        UUUS9$ )	N r   r   rq   r   rB   c              3   .   #    U  H  oc  M  Uv   M     g 7fr%   r   ).0vs     r   	<genexpr>&GLPNEncoder.forward.<locals>.<genexpr>l  s     m$[q$[s   	last_hidden_stater+   
attentions)r   	enumeratezipr   r   rJ   rw   rv   r}   tupler   )r(   rR   r   output_hidden_statesreturn_dictall_hidden_statesall_self_attentionsr   r+   idxr   embedding_layerblock_layer
norm_layerrU   rV   r   blklayer_outputss                      r   r-   GLPNEncoder.forwardM  s3    #7BD$5b4!''*
$D$9$94::t WXFC784O*+:=+I(M5#K03 #M5 T -a 0$$*=qAQ@S*S'	 1 '}5M)11*eRPXXYZ\]_`bcdooqM##$58H$H! Y" m]GZ$[mmm++*
 	
r    )r   rm   rJ   r   )FFTr   r=   s   @r   r   r     s    .
f  "$
 $
r    r   c                   0    \ rS rSr% \\S'   SrSrSr/ r	Sr
g)GLPNPreTrainedModelit  rm   glpnrR   )imager   N)r3   r4   r5   r6   r   __annotations__base_model_prefixmain_input_nameinput_modalities_no_split_modulesr;   r   r    r   r  r  t  s    $O!r    r  c                      ^  \ rS rSrU 4S jr\   SS\R                  S\S-  S\S-  S\S-  S\	\
-  4
S	 jj5       rS
rU =r$ )	GLPNModeli}  c                 p   > [         TU ]  U5        Xl        [        U5      U l        U R                  5         g r%   )r&   r'   rm   r   encoder	post_initr(   rm   r)   s     r   r'   GLPNModel.__init__  s/      #6* 	r    NrR   r   r   r   r   c                 0   Ub  UOU R                   R                  nUb  UOU R                   R                  nUb  UOU R                   R                  nU R	                  UUUUS9nUS   nU(       d	  U4USS  -   $ [        UUR                  UR                  S9$ )Nr   r   r   r   r   r   )rm   r   r   r   r  r   r+   r   )r(   rR   r   r   r   kwargsencoder_outputssequence_outputs           r   r-   GLPNModel.forward  s     2C1N-TXT_T_TqTq$8$D $++JjJj 	 &1%<k$++BYBY,,/!5#	 ' 
 *!,#%(;;;-)77&11
 	
r    )rm   r  )NNN)r3   r4   r5   r6   r'   r	   r   FloatTensorboolr   r   r-   r;   r<   r=   s   @r   r  r  }  sn     
 *.,0#'
''
  $;
 #Tk	

 D[
 
	 
 
r    r  c                   6   ^  \ rS rSrSrSU 4S jjrS rSrU =r$ )GLPNSelectiveFeatureFusioni  z
Selective Feature Fusion module, as explained in the [paper](https://huggingface.co/papers/2201.07436) (section 3.4). This
module adaptively selects and integrates local and global features by attaining an attention map for each feature.
c           
      X  > [         TU ]  5         [        R                  " [        R                  " [        US-  5      USSSS9[        R                  " U5      [        R                  " 5       5      U l        [        R                  " [        R                  " U[        US-  5      SSSS9[        R                  " [        US-  5      5      [        R                  " 5       5      U l	        [        R                  " [        US-  5      SSSSS9U l
        [        R                  " 5       U l        g )NrB   r   r   )in_channelsout_channelsrD   rE   rF   )r&   r'   r   
SequentialrG   r`   BatchNorm2dReLUconvolutional_layer1convolutional_layer2convolutional_layer3Sigmoidsigmoid)r(   
in_channelr)   s     r   r'   #GLPNSelectiveFeatureFusion.__init__  s    $&MMII#j1n"5J\]fgqrsNN:&GGI%
! %'MMII*3zA~;N\]fgqrsNN3zA~./GGI%
! %'IIJN+!ST^_%
! zz|r    c                 <   [         R                  " X4SS9nU R                  U5      nU R                  U5      nU R	                  U5      nU R                  U5      nXS S 2SS S 2S S 24   R                  S5      -  X$S S 2SS S 2S S 24   R                  S5      -  -   nU$ )Nr   rs   r   )r   catr"  r#  r$  r&  	unsqueeze)r(   local_featuresglobal_featuresfeaturesattnhybrid_featuress         r   r-   "GLPNSelectiveFeatureFusion.forward  s    99n>AF,,X6,,X6,,X6||H%(1a
+;+E+Ea+HH?q!QJ^

)A,L  r    )r"  r#  r$  r&  )@   rX   r=   s   @r   r  r    s    
$* r    r  c                   2   ^  \ rS rSrU 4S jrSS jrSrU =r$ )GLPNDecoderStagei  c                    > [         TU ]  5         X:H  nU(       d  [        R                  " XSS9O[        R                  " 5       U l        [        U5      U l        [        R                  " SSSS9U l	        g )Nr   )rD   rB   bilinearFscale_factormodealign_corners)
r&   r'   r   rG   r   convolutionr  fusionUpsampleupsample)r(   r  r  should_skipr)   s       r   r'   GLPNDecoderStage.__init__  sX    !1Va299[ANgigrgrgt0>SXYr    c                 r    U R                  U5      nUb  U R                  X5      nU R                  U5      nU$ r%   r;  r<  r>  )r(   hidden_stateresiduals      r   r-   GLPNDecoderStage.forward  s:    ''5;;|>L}}\2r    rB  r%   r   r=   s   @r   r4  r4    s    Z	 	r    r4  c                   n   ^  \ rS rSrU 4S jrS\\R                     S\\R                     4S jrSr	U =r
$ )GLPNDecoderi  c           	      0  > [         TU ]  5         UR                  S S S2   nUR                  n[        R
                  " U Vs/ s H  n[        XC5      PM     sn5      U l        S U R                  S   l        [        R                  " SSSS9U l
        g s  snf )Nrq   r   rB   r6  Fr7  )r&   r'   r   decoder_hidden_sizer   r   r4  stagesr<  r=  final_upsample)r(   rm   reserved_hidden_sizesr  rM   r)   s        r   r'   GLPNDecoder.__init__  s     & 3 3DbD 911mmLabLa[k8Lab
 !%A kkqzY^_ cs   Br+   r   c                     / nS n[        US S S2   U R                  5       H  u  pEU" XC5      nUR                  U5        M      U R                  U5      US'   U$ )Nrq   )r   rJ  r   rK  )r(   r+   stage_hidden_statesstage_hidden_staterC  stages         r   r-   GLPNDecoder.forward  si     !#&}TrT':DKK#HL!&|!H&&'9: $I #'"5"56H"IB""r    )rK  rJ  r3   r4   r5   r6   r'   listr   r9   r-   r;   r<   r=   s   @r   rG  rG    s3    `	#T%,,%7 	#D<N 	# 	#r    rG  c                   6   ^  \ rS rSrSrSU 4S jjrS rSrU =r$ )	SiLogLossi  z
Implements the Scale-invariant log scale loss [Eigen et al., 2014](https://huggingface.co/papers/1406.2283).

$$L=\frac{1}{n} \sum_{i} d_{i}^{2}-\frac{1}{2 n^{2}}\left(\sum_{i} d_{i}^{2}\right)$$ where $d_{i}=\log y_{i}-\log
y_{i}^{*}$.

c                 .   > [         TU ]  5         Xl        g r%   )r&   r'   lambd)r(   rX  r)   s     r   r'   SiLogLoss.__init__  s    
r    c                 f   US:  R                  5       n[        R                  " X#   5      [        R                  " X   5      -
  n[        R                  " [        R                  " US5      R                  5       U R                  [        R                  " UR                  5       S5      -  -
  5      nU$ )Nr   rB   )detachr   logrz   powmeanrX  )r(   predtarget
valid_maskdiff_loglosss         r   r-   SiLogLoss.forward  s    qj((*
99V/0599T=M3NNzz%))Ha0557$**uyyQYQ^Q^Q`bcGd:dder    )rX  )g      ?rX   r=   s   @r   rV  rV    s     r    rV  c                   h   ^  \ rS rSrU 4S jrS\\R                     S\R                  4S jrSr	U =r
$ )GLPNDepthEstimationHeadi  c                    > [         TU ]  5         Xl        UR                  n[        R
                  " [        R                  " X"SSSS9[        R                  " SS9[        R                  " USSSSS95      U l        g )Nr   r   rC   F)inplace)	r&   r'   rm   rI  r   r  rG   r!  head)r(   rm   channelsr)   s      r   r'    GLPNDepthEstimationHead.__init__  s`    --MMIIha1MGGE"IIhqAF
	r    r+   r   c                     XR                   R                     nU R                  U5      n[        R                  " U5      U R                   R
                  -  nUR                  SS9nU$ )Nr   rs   )rm   head_in_indexri  r   r&  	max_depthsqueeze)r(   r+   predicted_depths      r   r-   GLPNDepthEstimationHead.forward)  sX    %kk&?&?@		-0--69N9NN)11a18r    )rm   ri  rS  r=   s   @r   rf  rf    s-    

	T%,,%7 	ELL 	 	r    rf  zg
    GLPN Model transformer with a lightweight depth estimation head on top e.g. for KITTI, NYUv2.
    )custom_introc                      ^  \ rS rSrU 4S jr\    SS\R                  S\R                  S-  S\S-  S\S-  S\S-  S	\	\R                     \-  4S
 jj5       rSrU =r$ )GLPNForDepthEstimationi5  c                    > [         TU ]  U5        [        U5      U l        [	        U5      U l        [        U5      U l        U R                  5         g r%   )	r&   r'   r  r  rG  decoderrf  ri  r  r  s     r   r'   GLPNForDepthEstimation.__init__;  s@     f%	"6*+F3	 	r    NrR   labelsr   r   r   r   c                    Ub  UOU R                   R                  nUb  UOU R                   R                  nU R                  UUSUS9nU(       a  UR                  OUS   nU R                  U5      n	U R                  U	5      n
SnUb  [        5       nU" X5      nU(       d%  U(       a
  U
4USS -   nO	U
4USS -   nUb  U4U-   $ U$ [        UU
U(       a  UR                  OSUR                  S9$ )a  
labels (`torch.FloatTensor` of shape `(batch_size, height, width)`, *optional*):
    Ground truth depth estimation maps for computing the loss.

Examples:

```python
>>> from transformers import AutoImageProcessor, GLPNForDepthEstimation
>>> import torch
>>> import numpy as np
>>> from PIL import Image
>>> import httpx
>>> from io import BytesIO

>>> url = "http://images.cocodataset.org/val2017/000000039769.jpg"
>>> with httpx.stream("GET", url) as response:
...     image = Image.open(BytesIO(response.read()))

>>> image_processor = AutoImageProcessor.from_pretrained("vinvino02/glpn-kitti")
>>> model = GLPNForDepthEstimation.from_pretrained("vinvino02/glpn-kitti")

>>> # prepare image for the model
>>> inputs = image_processor(images=image, return_tensors="pt")

>>> with torch.no_grad():
...     outputs = model(**inputs)

>>> # interpolate to original size
>>> post_processed_output = image_processor.post_process_depth_estimation(
...     outputs,
...     target_sizes=[(image.height, image.width)],
... )

>>> # visualize the prediction
>>> predicted_depth = post_processed_output[0]["predicted_depth"]
>>> depth = predicted_depth * 255 / predicted_depth.max()
>>> depth = depth.detach().cpu().numpy()
>>> depth = Image.fromarray(depth.astype("uint8"))
```NTr  r   rB   )rc  rp  r+   r   )
rm   r   r   r  r+   rv  ri  rV  r   r   )r(   rR   rx  r   r   r   r  r   r+   outrp  rc  loss_fctr   s                 r   r-   GLPNForDepthEstimation.forwardE  s   b &1%<k$++BYBY$8$D $++JjJj 	 ))/!%#	  
 2=--'!*ll=)))C. {HO4D#)+gabk9)+gabk9)-)9TGf$EvE#+3G'//T))	
 	
r    )rv  r  ri  )NNNN)r3   r4   r5   r6   r'   r	   r   r  r  r   r9   r   r-   r;   r<   r=   s   @r   rt  rt  5  s      ,0)-,0#'R
''R
 !!D(R
  $;	R

 #TkR
 D[R
 
u||	3	3R
 R
r    rt  )rt  r   r  r  )r   F)*r7   ry   r   r   activationsr   modeling_outputsr   r   modeling_utilsr   utilsr	   r
   configuration_glpnr   
get_loggerr3   loggerr9   r8   r  r   Moduler"   r?   rZ   r   r   r   r   r   r   r  r  r  r4  rG  rV  rf  rt  __all__r   r    r   <module>r     s       ! E - , * 
		H	%U\\ e T V[VbVb  %299 %) )4L L`	RYY 	BII (  0(		 (VU
")) U
p /   +
# +
 +
\) )Xryy (#")) #6		 *bii 2 
^
0 ^

^
B Vr    