jddlZddlZddlmZddlmZddlZddlm Z ddl m Z gdZ Gdd e ZGd d e ZGd d e ZGdde ZGdde ZGdde ZGdde ZdS)N)Callable)Any) NestedObject) functional) SampleComposeImageTransformColorInversionOneOf RandomApply RandomRotate RandomCropceZdZUdZdgZeeed<deee e ge e e ffddfdZ de d e de e e ffd Z dS) raImplements a wrapper that will apply transformations sequentially on both image and target .. code:: python >>> import numpy as np >>> import torch >>> from doctr.transforms import SampleCompose, ImageTransform, ColorInversion, RandomRotate >>> transfos = SampleCompose([ImageTransform(ColorInversion((32, 32))), RandomRotate(30)]) >>> out, out_boxes = transfos(torch.rand(8, 64, 64, 3), np.zeros((2, 4))) Args: transforms: list of transformation modules sample_transforms_children_names transformsreturnNc||_dSNrselfrs `/var/www/html/Carbon-Document/venv/lib/python3.11/site-packages/doctr/transforms/modules/base.py__init__zSampleCompose.__init__%s!+xtargetc<|jD]}|||\}}||fSrr)rrrts r__call__zSampleCompose.__call__(s3' % %A!V IAvv&yr) __name__ __module__ __qualname____doc__rliststr__annotations__rrtuplerr rrrrs  #6!6OT#Y666,4#sU38_1L(M#N,SW,,,,#suS#XrrcveZdZUdZdgZeeed<dee ge fddfdZ de d e de e e ffd Z dS) r aImplements a transform wrapper to turn an image-only transformation into an image+target transform .. code:: python >>> import torch >>> from doctr.transforms import ImageTransform, ColorInversion >>> transfo = ImageTransform(ColorInversion((32, 32))) >>> out, _ = transfo(torch.rand(8, 64, 64, 3), None) Args: transform: the image transformation module to wrap img_transformr transformrNc||_dSrr+)rr,s rrzImageTransform.__init__?s&rimgrc4||}||fSrr.rr/rs rr zImageTransform.__call__Bs  %%F{r) r!r"r#r$rr%r&r'rrrr(r r)rrr r /s  #2!2OT#Y222'(C5#:"6'4''''CsCxrr c@eZdZdZd deddfdZdefdZdedefd Z dS) r aApplies the following tranformation to a tensor (image or batch of images): convert to grayscale, colorize (shift 0-values randomly), and then invert colors .. code:: python >>> import torch >>> from doctr.transforms import ColorInversion >>> transfo = ColorInversion(min_val=0.6) >>> out = transfo(torch.rand(8, 64, 64, 3)) Args: min_val: range [min_val, 1] to colorize RGB pixels ?min_valrNc||_dSrr4)rr4s rrzColorInversion.__init__Vs  rcd|jS)Nzmin_val=r6rs r extra_reprzColorInversion.extra_reprYs($,(((rr/c6tj||jSr)F invert_colorsr4)rr/s rr zColorInversion.__call__\ssDL111rr3) r!r"r#r$floatrr&r9rr r)rrr r Gs  )C))))2C2C222222rr c eZdZUdZdgZeeed<deee ge fddfdZ d de de j dzde e e e j fzfd ZdS) r aMRandomly apply one of the input transformations .. code:: python >>> import torch >>> from doctr.transforms import OneOf >>> transfo = OneOf([JpegQuality(), Gamma()]) >>> out = transfo(torch.rand(1, 64, 64, 3)) Args: transforms: list of transformations, one only will be picked rrrNc||_dSr)rrs rrzOneOf.__init__ps $rr/rc|jttjt|jz}| ||n |||Sr)rintrandomlen)rr/rtransfos rr zOneOf.__call__ssM/#fmooDO8L8L&L"M"MN%~wws|||7733G3GGrr)r!r"r#r$rr%r&r'rrrnpndarrayr(r r)rrr r `s  #/OT#Y///%4#(<#=%$%%%%HHCHd):HcERUWYWaRaLbFbHHHHHHrr c eZdZdZd deegefdeddfdZdefdZ d d ed e j dzdee ee j fzfd Z dS)r aXApply with a probability p the input transformation .. code:: python >>> import torch >>> from doctr.transforms import RandomApply >>> transfo = RandomApply(Gamma(), p=.5) >>> out = transfo(torch.rand(1, 64, 64, 3)) Args: transform: transformation to apply p: probability to apply r3r,prNc"||_||_dSrr,rI)rr,rIs rrzRandomApply.__init__s"rc&d|jd|jS)Nz transform=z, p=rKr8s rr9zRandomApply.extra_reprs8DN88888rr/rctj|jkr-|||n|||S||n||fSr)rCrIr,r1s rr zRandomApply.__call__sQ =??TV # #*0.4>>#&&&dnnSRX>Y>Y Ynss3-7rr=r)r!r"r#r$rrr>rr&r9rFrGr(r r)rrr r zs  (C5#:"6549C999988C8d):8cERUWYWaRaLbFb888888rr cleZdZdZd dededdfdZdefd Zd e d e j de e e j ffd Z dS)r aqRandomly rotate a tensor image and its boxes .. image:: https://doctr-static.mindee.com/models?id=v0.4.0/rotation_illustration.png&src=0 :align: center Args: max_angle: maximum angle for rotation, in degrees. Angles will be uniformly picked in [-max_angle, max_angle] expand: whether the image should be padded before the rotation @F max_angleexpandrNc"||_||_dSrrPrQ)rrPrQs rrzRandomRotate.__init__s" rc&d|jd|jS)Nz max_angle=z , expand=rSr8s rr9zRandomRotate.extra_reprsBDNBBT[BBBrr/rctj|j |j}tj||||j\}}|d|dkddk}|||fS)Nr) rCuniformrPr; rotate_samplerQmaxminsum)rr/rangler_imgr_polysis_kepts rr zRandomRotate.__call__su??feT[IIw;;q>>GKKNN277::a?gg&&&r)rOF)r!r"r#r$r>boolrr&r9rrFrGr(r r)rrr r s%tCCCCCC'C''c2:o8N''''''rr ceZdZdZd deeefdeeefddfdZdefd Zd e d e j dee e j ffd Z dS)rzRandomly crop a tensor image and its boxes Args: scale: tuple of floats, relative (min_area, max_area) of the crop ratio: tuple of float, relative (min_ratio, max_ratio) where ratio = h/w g{Gz?g?g?gHzG?scaleratiorNc"||_||_dSrrdre)rrdres rrzRandomCrop.__init__s  rc&d|jd|jS)Nzscale=z, ratio=rgr8s rr9zRandomCrop.extra_reprs8 88DJ888rr/rctj|jd|jd}tj|jd|jd}|jdd\}}||z|z}|||z z}t t tj||z} t t tj||z } t| |} t| |} tj d|| z } tj d|| z } | |z | |z | | z|z | | z|z f} |jdddkrEtj |d}tj |d}tj ||fd}n|}tj||| \}}|jddkr||fS|tj|ddfS)NrrV)r)axis)rCrWrdreshaperBroundmathsqrtrZrandintrFrY concatenater;crop_detectionclip)rr/rrdreheightwidth crop_area aspect_ratio crop_width crop_heightrycrop_boxmin_xymax_xy_target croped_img crop_boxess rr zRandomCrop.__call__stz!}djm<<tz!}djm<< "## EMF* / ty\)ABBCCDD % )l*B C CDDEE U++ +v..  N1ej0 1 1 N1f{2 3 3Iq6zA Ne+Ca+oY_E_` < v % %VF+++FVF+++Fnff%5A>>>GGG"#!1#w!I!I J  A ! # #; 27:q!4444r)rbrc) r!r"r#r$r(r>rr&r9rrFrGr r)rrrrseE5L1eUZlH[os9C9999#5C#5#5c2:o8N#5#5#5#5#5#5rr)rorCcollections.abcrtypingrnumpyrFdoctr.utils.reprrrr;__all__rr r r r r rr)rrrs  $$$$$$)))))) u u uL6\022222\2222HHHHHLHHH488888,8888'''''<'''425252525252525252525r