j_ UddlZddlmZddlmZddlmZddlmZ ddl m Z ddl m Z d gZd d d d ee d ddiZeeeeeffed< ddededededeedzdede jfdZddedede jfdZdS)N)deepcopy)Any)nn)vgg)VOCABSload_pretrained_params vgg16_bn_r)gh|?5?g= ףp=?gV-?)gA`"?gl?g$C?)r r frenchzMhttps://doctr-static.mindee.com/models?id=v0.4.1/vgg16_bn_r-d108c19c.pt&src=0)meanstd input_shapeclassesurl default_cfgsarch pretrainedtv_archnum_rect_pools ignore_keyskwargsreturnc |dtt|d|d<|dt|d|d<tt|}|d|d<|d|d<|dt j|di|ddi}dt|jD}|| dD]} tj d|j| <tj d|_ tj d|d|_dtd t d dfd } t#j| ||_|rV|dtt|dkr|nd} |t|d | ||_|S)N num_classesrweightscLg|]!\}}t|tj|"S) isinstancer MaxPool2d).0idxms j/var/www/html/Carbon-Document/venv/lib/python3.11/site-packages/doctr/models/classification/vgg/pytorch.py z_vgg..2s-\\\a 1bl@[@[\\\\))r)r)i path_or_urlrrc"t||fi|dS)zLoad pretrained parameters onto the model Args: path_or_url: the path or URL to the model parameters (checkpoint) **kwargs: additional arguments to be passed to `doctr.models.utils.load_pretrained_params` Nr )selfr*rs r%from_pretrainedz_vgg..from_pretrained;s" t[;;F;;;;;r'r)rr)getlenrrpoptv_vgg__dict__ enumeratefeaturesrr!AdaptiveAvgPool2davgpoolLinear classifierstrrtypes MethodTyper-cfg) rrrrrr_cfgmodel pool_idcsr#r- _ignore_keyss r%_vggrAs#JJ}c,t:LY:W6X6XYYF= 9l4.@.KLLF9 L& ' 'D /DY'DO JJy OG $ < `_, modified by adding batch normalization, rectangular pooling and a simpler classification head. >>> import torch >>> from doctr.models import vgg16_bn_r >>> model = vgg16_bn_r(pretrained=False) >>> input_tensor = torch.rand((1, 3, 512, 512), dtype=torch.float32) >>> out = model(input_tensor) Args: pretrained (bool): If True, returns a model pre-trained on ImageNet **kwargs: keyword arguments of the VGG architecture Returns: VGG feature extractor r vgg16_bnrrzclassifier.weightzclassifier.bias)rA)rrs r%r r SsA$     )*;<      r')rN)F)r:copyrtypingrtorchrtorchvision.modelsrr1doctr.datasetsrutilsr __all__listrdictr9__annotations__boolintVGGrAr rr'r%rQs  ,,,,,,!!!!!!++++++ .%$"4x())^ + d3S#X&'$( 11 111 1 cT! 1  1 Z1111h436:r'