QЦi,XSSKrSSKJrJr SSKrSSKrSSKJrJr SSKJ r SSK J r SSK J r SSKJrJr SS KJr S S KJrJr S rS rSr\"SS9S5r\R4R6SS\S\\\\4S\ \S\S\S\S\4Sjj5r"SS\R@5r!g)N)ListUnion)nnTensor)is_compile_supported)BroadcastingList2)_pair)_assert_has_ops_has_ops)_log_api_usage_once)check_roi_boxes_shapeconvert_boxes_to_roi_formatc ^U4SjnU$)zcLazily wrap a function with torch.compile on the first call This avoids eagerly importing dynamo. cJ>^[R"T5UU4Sj5nU$)Nc>[R"T40TD6n[R"T5"U5[ 5TR 'U"U0UD6$N)torchcompile functoolswrapsglobals__name__)argskwargs compiled_fncompile_kwargsfns X/var/www/html/ai-image-ml/venv/lib/python3.13/site-packages/torchvision/ops/roi_align.py compile_hook7lazy_compile..decorate_fn..compile_hooksD--=n=K%.__R%8%EGIbkk "// /)rr)rr!rs` r decorate_fn!lazy_compile..decorate_fns&   0  0 r#)rr$s` r lazy_compiler's  r#c^^^^^TR5unmpxURSS9nURSS9nUR5n UR5n [R"XS- :US- U S-5n [R"XS- :US- U 5n [R"XS- :UR TR 5U5n[R"XS- :US- U S-5n [R"XS- :US- U 5n [R"XS- :UR TR 5U5nX)- n X:- nSU - nSU- nUUUUU4SjnU"X5nU"X5nU"X5nU"X5nSnU"UU5nU"X5nU"U U5nU"X5nUU-UU--UU--UU--nU$)Nrminr?c P>TbITce[R"TSS2SSS24US5n[R"TSS2SSS24US5nTTSS2SSSSS4[R"TTRS9SSS2SSSS4USS2SSS2SSS2S4USS2SSSS2SSS244$)Nrdevice)rwherearanger.)yxchannelsinput roi_batch_indxmaskymasks r masked_index+_bilinear_interpolate..masked_indexBs  $ $$ E!T1*-q!4A E!T1*-q!4A !T4tT9 : LL%,, 7atTSW8W X aq$4' ( atQa' ( *  r#c HUSS2SSS2SSS2S4USS2SSSS2SSS24-$rr&)r1r2s r outer_prod)_bilinear_interpolate..outer_prodWs1D!T1d*+a4q$0I.JJJr#)sizeclampintrr/todtype)r4r5r1r2r7r6_heightwidthy_lowx_lowy_highx_highlylxhyhxr8v1v2v3v4r;w1w2w3w4valr3s`` `` @r _bilinear_interpolaterV#s"'Ax AA AA EEGE EEGE [[1*,fqj%!) DF KK!+VaZ ?E EaZ'ekk):A>A [[!)+UQY BF KK*EAIu =E EQY&U[[(91=A B B rB rB    e #B e $B f $B f %BK B B B B B B B B r'BG b2g %R /C Jr#c[R"5(a?UR(a.UR[R:waUR 5$U$r)ris_autocast_enabledis_cudarAdoublefloat)tensors r maybe_castr]es7   ""v~~&,,%,,:V||~ r#T)dynamicc URn[U5n[U5nUR5u pn [R"X0R S9n [R"X@R S9n USS2S4R 5n U(aSOSnUSS2S4U-U- nUSS2S4U-U- nUSS2S4U-U- nUSS2S4U-U- nUU- nUU- nU(d*[R"US S 9n[R"US S 9nUU- nUU- nUS:nU(aUO[R"UU- 5nU(aUO[R"UU- 5nU(aR[UU-S5n[R"UUR S9n[R"UUR S9nSnSnOz[R"UU-SS 9n[R"XR S9n[R"XR S9nUSSS24USS2S4:nUSSS24USS2S4:nS nU"U5U SSS2S4U"U5--USSSS24S-RUR5U"UU- 5--n U"U5U SSS2S4U"U5--USSSS24S-RUR5U"UU- 5--n![X U U!UU5n"U(dJ[R"USS2SSSSS2S4U"S5n"[R"USS2SSSSSS24U"S5n"U"RS 5n#[U[R5(aU#USS2SSS4-n#OU#U-n#U#RU5n#U#$) Nr-rg?grr r+r)cUSS2SS4$rr&)ts r from_K_roi_align..from_KsD$r#))rAr]r=rr0r.r?r>ceilmaxr@rVr/sum isinstancer)$r4rois spatial_scale pooled_height pooled_widthsampling_ratioaligned orig_dtyperBrCrDphpwr5offset roi_start_w roi_start_h roi_end_w roi_end_h roi_width roi_height bin_size_h bin_size_wexact_samplingroi_bin_grid_hroi_bin_grid_wcountiyixr7r6rdr1r2rUoutputs$ r _roi_alignrrsJ u E d D**,Aq% mLL 9B l<< 8B AJNN$MSFq!t*},v5Kq!t*},v5KQT ]*V3IQT ]*V3IK'I[(J KK s3 [[5 m+J\)J#a'N'5^5::jS`F`;aN'5^5::iR^F^;_NN^3Q7 \\. > \\. > N^;C \\& 6 \\% 547 nQW5547 nQW55  { T1d] fZ0 0 1 dD!m s " & &u{{ 3fZ.=X6Y Y Z { T1d] fZ0 0 1 dD!m s " & &u{{ 3fZ.=X6Y Y Z aE5 IC kk%4tQ <=sAFkk%4tT1 <=sAF WWX F%&&%4t+,,% YYz "F Mr#r4boxes output_sizermrprqreturnc [RR5(d2[RR5(d[ [ 5 [ U5 Un[U5n[U[R5(d [U5n[RR5(d[5(a<[R"5(aZUR(dUR(a8[UR R"5(a[%XX2SUSXE5$['5 [R(R*R XX2SUSXE5$)a Performs Region of Interest (RoI) Align operator with average pooling, as described in Mask R-CNN. Args: input (Tensor[N, C, H, W]): The input tensor, i.e. a batch with ``N`` elements. Each element contains ``C`` feature maps of dimensions ``H x W``. If the tensor is quantized, we expect a batch size of ``N == 1``. boxes (Tensor[K, 5] or List[Tensor[L, 4]]): the box coordinates in (x1, y1, x2, y2) format where the regions will be taken from. The coordinate must satisfy ``0 <= x1 < x2`` and ``0 <= y1 < y2``. If a single Tensor is passed, then the first column should contain the index of the corresponding element in the batch, i.e. a number in ``[0, N - 1]``. If a list of Tensors is passed, then each Tensor will correspond to the boxes for an element i in the batch. output_size (int or Tuple[int, int]): the size of the output (in bins or pixels) after the pooling is performed, as (height, width). spatial_scale (float): a scaling factor that maps the box coordinates to the input coordinates. For example, if your boxes are defined on the scale of a 224x224 image and your input is a 112x112 feature map (resulting from a 0.5x scaling of the original image), you'll want to set this to 0.5. Default: 1.0 sampling_ratio (int): number of sampling points in the interpolation grid used to compute the output value of each pooled output bin. If > 0, then exactly ``sampling_ratio x sampling_ratio`` sampling points per bin are used. If <= 0, then an adaptive number of grid points are used (computed as ``ceil(roi_width / output_width)``, and likewise for height). Default: -1 aligned (bool): If False, use the legacy implementation. If True, pixel shift the box coordinates it by -0.5 for a better alignment with the two neighboring pixel indices. This version is used in Detectron2 Returns: Tensor[K, C, output_size[0], output_size[1]]: The pooled RoIs. rr)rjit is_scripting is_tracingr roi_alignrr rkrrr $are_deterministic_algorithms_enabledrYis_mpsrr.typerr ops torchvision)r4rrrmrprqrls r rrsR 99 ! ! # #EII,@,@,B,BI&% D $K dELL ) )*40 99 ! ! # # uIIKKQVQ^Q^bgbnbn"5<<#4#455e=a.+VW.Zhr r 99 * * ]NKNN r#c ^\rSrSrSrSS\\S\S\S\4U4Sjjjr S\ S \ \ \ \ 4S \ 4S jr S \4S jrS rU=r$)RoIAligniz See :func:`roi_align`. rrmrprqch>[TU]5 [U5 XlX lX0lX@lgr)super__init__r rrmrprq)selfrrmrprq __class__s r rRoIAlign.__init__ s/ D!&*, r#r4rlrcp[XURURURUR5$r)rrrmrprq)rr4rls r forwardRoIAlign.forwards.d&6&68J8JDL_L_aeamamnnr#c URRSURSURSURSUR S3 nU$)Nz (output_size=z, spatial_scale=z, sampling_ratio=z , aligned=))rrrrmrprq)rss r __repr__RoIAlign.__repr__s^~~&&'(++,t112 3 34   r#)rqrrprm)F)r __module__ __qualname____firstlineno____doc__rr?r[boolrrrrrstrr__static_attributes__ __classcell__)rs@r rrs &s+      oVo5f1E+Fo6o #  r#r)r+rfF)"rtypingrrrtorch.fxrrtorch._dynamo.utilsrtorch.jit.annotationsrtorch.nn.modules.utilsr torchvision.extensionr r utilsr _utilsrrr'rVr]rfxwrapr?r[rrModulerr&r#r rs 43(;'F&=DdUUp  7 7 f% &7#3'7 7  7  7 77tryyr#