sghVdZddlZddlmZmZmZmZmZddlZ ddl m Z m Z m Z ddlmZmZmZddlmZmZmZmZmZmZmZmZmZmZmZmZddlm Z m!Z!m"Z"m#Z#m$Z$m%Z%m&Z&e#rddl'Z'e$rddl(Z(e"rdd l)m*Z+e%jXe-Z.d Z/d Z0d Z1d Z2dZ3Gdde Z4y)z Image processor class for OWLv2.N)DictListOptionalTupleUnion)BaseImageProcessor BatchFeature get_size_dict)center_to_corners_formatpadto_channel_dimension_format) OPENAI_CLIP_MEANOPENAI_CLIP_STDChannelDimension ImageInputPILImageResamplingget_image_sizeinfer_channel_dimension_formatis_scaled_imagemake_list_of_imagesto_numpy_array valid_imagesvalidate_preprocess_arguments) TensorTypefilter_out_non_signature_kwargsis_scipy_availableis_torch_availableis_vision_availableloggingrequires_backends)ndimagec|jr>|jtjtjfvr|S|j S|jtj tjfvr|S|jS)N) is_floating_pointdtypetorchfloat32float64floatint32int64int)ts c/var/www/html/venv/lib/python3.12/site-packages/transformers/models/owlv2/image_processing_owlv2.py_upcastr/Ds`GG u}}==qL1779LGG U[[99qFquuwFcft|}|dddf|dddfz |dddf|dddfz zS)a Computes the area of a set of bounding boxes, which are specified by its (x1, y1, x2, y2) coordinates. Args: boxes (`torch.FloatTensor` of shape `(number_of_boxes, 4)`): Boxes for which the area will be computed. They are expected to be in (x1, y1, x2, y2) format with `0 <= x1 < x2` and `0 <= y1 < y2`. Returns: `torch.FloatTensor`: a tensor containing the area for each box. Nrr)r/)boxess r.box_arear5MsB ENE !Q$K%1+ %%1+ad *C DDr0c^t|}t|}tj|dddddf|ddddf}tj|dddddf|ddddf}||z j d}|dddddf|dddddfz}|dddf|z|z }||z } | |fS)Nr2r)minr3)r5r&maxr7clamp) boxes1boxes2area1area2left_top right_bottom width_heightinterunionious r.box_iourD]s V E V Eyy4! ,fQUm input.ndim`. output_shape (`Tuple`): The output shape converted to tuple. Raises ------ ValueError: If output_shape length is smaller than the image number of dimensions. Notes ----- The input image is reshaped if its number of dimensions is not equal to output_shape_length. )r3r3zIoutput_shape length cannot be smaller than the image number of dimensions)tuplelenshapendimnpreshape ValueError)image output_shape output_ndim input_shapes r._preprocess_resize_output_shaperRms.&Ll#K++KUZZt{UZZ788  5+. ,   Q &#u{{2&88  ,  uzz !ghh , r0c*tj|}tj|r)tj}tj}||}n tj}tj }||}tj |||}|S)aClip output image to range of values of input image. Note that this function modifies the values of *output_image* in-place. Taken from: https://github.com/scikit-image/scikit-image/blob/b4b521d6f0a105aabeaa31699949f78453ca3511/skimage/transform/_warps.py#L640. Args: input_image : ndarray Input image. output_image : ndarray Output image, which is modified in-place. )rKr7isnannanminnanmaxr8clip) input_image output_imagemin_valmin_funcmax_funcmax_vals r._clip_warp_outputr^snff[!G xx9999;'6666{#G77<':L r0ceZdZdZdgZdddddej dddf dedee e fded ed e e e fd ed ed e ee ee fde ee ee fddffd Z ddej"de ee efde ee effdZ ddej(d e e e fdede ee efde ee efdej(f dZedddddddddej.df deded ed e e e fdede d ed e ee ee fde ee ee fde ee efdede ee efdej6j6fdZ d de deeeeffdZd!dZxZ S)"Owlv2ImageProcessorav Constructs an OWLv2 image processor. Args: do_rescale (`bool`, *optional*, defaults to `True`): Whether to rescale the image by the specified scale `rescale_factor`. Can be overriden by `do_rescale` in the `preprocess` method. rescale_factor (`int` or `float`, *optional*, defaults to `1/255`): Scale factor to use if rescaling the image. Can be overriden by `rescale_factor` in the `preprocess` method. do_pad (`bool`, *optional*, defaults to `True`): Whether to pad the image to a square with gray pixels on the bottom and the right. Can be overriden by `do_pad` in the `preprocess` method. do_resize (`bool`, *optional*, defaults to `True`): Controls whether to resize the image's (height, width) dimensions to the specified `size`. Can be overriden by `do_resize` in the `preprocess` method. size (`Dict[str, int]` *optional*, defaults to `{"height": 960, "width": 960}`): Size to resize the image to. Can be overriden by `size` in the `preprocess` method. resample (`PILImageResampling`, *optional*, defaults to `Resampling.BILINEAR`): Resampling method to use if resizing the image. Can be overriden by `resample` in the `preprocess` method. do_normalize (`bool`, *optional*, defaults to `True`): Whether to normalize the image. Can be overridden by the `do_normalize` parameter in the `preprocess` method. image_mean (`float` or `List[float]`, *optional*, defaults to `OPENAI_CLIP_MEAN`): Mean to use if normalizing the image. This is a float or list of floats the length of the number of channels in the image. Can be overridden by the `image_mean` parameter in the `preprocess` method. image_std (`float` or `List[float]`, *optional*, defaults to `OPENAI_CLIP_STD`): Standard deviation to use if normalizing the image. This is a float or list of floats the length of the number of channels in the image. Can be overridden by the `image_std` parameter in the `preprocess` method. pixel_valuesTgp?N do_rescalerescale_factordo_pad do_resizesizeresample do_normalize image_mean image_stdreturnc  t |di| ||_||_||_||_||nddd|_||_||_||nt|_ | | |_ yt|_ y)Ni)heightwidth) super__init__rbrcrdrerfrgrhrrirrj) selfrbrcrdrerfrgrhrirjkwargs __class__s r.rqzOwlv2ImageProcessor.__init__su "6"$, " ,DS32O   ((2(>*DT&/&;r0rN data_formatinput_data_formatcrt|\}}t||}t|d||z fd||z ffd||}|S)ae Pad an image to a square with gray pixels on the bottom and the right, as per the original OWLv2 implementation. Args: image (`np.ndarray`): Image to pad. data_format (`str` or `ChannelDimension`, *optional*): The channel dimension format of the image. If not provided, it will be the same as the input image. input_data_format (`ChannelDimension` or `str`, *optional*): The channel dimension format of the input image. If not provided, it will be inferred from the input image. rg?)rNpaddingconstant_valuesrurv)rr8r )rrrNrurvrmrnrfs r.r zOwlv2ImageProcessor.padsU&'u- 65!'!TE\):;#/   r0 anti_aliasingc (t|d|d|df}t|tj}t ||\}}|j } t j| |} d} d} d} |r|t jd| dz dz }nt j|t j| z}t j|dkr tdt j|dkD| dkzrtjd tj ||| | }n|}| Dcgc]}d|z  }}tj"||| | | d }t%||}t||tj}|t|||}|S|}|Scc}w) a  Resize an image as per the original implementation. Args: image (`np.ndarray`): Image to resize. size (`Dict[str, int]`): Dictionary containing the height and width to resize the image to. anti_aliasing (`bool`, *optional*, defaults to `True`): Whether to apply anti-aliasing when downsampling the image. anti_aliasing_sigma (`float`, *optional*, defaults to `None`): Standard deviation for Gaussian kernel when downsampling the image. If `None`, it will be calculated automatically. data_format (`str` or `ChannelDimension`, *optional*): The channel dimension format of the image. If not provided, it will be the same as the input image. input_data_format (`ChannelDimension` or `str`, *optional*): The channel dimension format of the input image. If not provided, it will be inferred from the input image. scipyrmrnmirrorrr3r2zFAnti-aliasing standard deviation must be greater than or equal to zerozWAnti-aliasing standard deviation greater than zero but not down-sampling along all axes)cvalmodeT)orderrr~ grid_mode)r!rrLASTrRrIrKdividemaximum atleast_1d ones_likeanyrMwarningswarnndigaussian_filterzoomr^)rrrNrfrzanti_aliasing_sigmarurvrsrOrQfactorsndi_moder~rfilteredf zoom_factorsouts r.resizezOwlv2ImageProcessor.resize s: $(XW 6 +E3C3H3HI=e\R|kk ))K6 "*&(jjWq[A4E&F#&(mm4G&H2<vay I  $ 5Rc dF ^deUZdhhU>OhPeFe $    &7F $U^opF ou ej '{N_ `  '>BBU= f  s$GG4G G9GG threshold target_sizescN|j|j}}|"t|t|k7r tdt j |d}t j |j}|j}t|}|t|trMt j|D cgc]} | d c} } t j|D cgc]} | d c} } n|jd\} } t j | | } t j| | | | gdj|j } || dddddfz}g}t#|||D]3\}}}|||kD}|||kD}|||kD}|j%|||d5|Scc} wcc} w)a| Converts the raw output of [`OwlViTForObjectDetection`] into final bounding boxes in (top_left_x, top_left_y, bottom_right_x, bottom_right_y) format. Args: outputs ([`OwlViTObjectDetectionOutput`]): Raw outputs of the model. threshold (`float`, *optional*): Score threshold to keep object detection predictions. target_sizes (`torch.Tensor` or `List[Tuple[int, int]]`, *optional*): Tensor of shape `(batch_size, 2)` or list of tuples (`Tuple[int, int]`) containing the target size `(height, width)` of each image in the batch. If unset, predictions will not be resized. Returns: `List[Dict]`: A list of dictionaries, each dictionary containing the scores, labels and boxes for an image in the batch as predicted by the model. NTMake sure that you pass in as many target sizes as the batch dimension of the logitsrFdimrr3scoreslabelsr4)logits pred_boxesrHrMr&r8sigmoidvaluesindicesr isinstancerTensorunbindstacktodevicezipappend)rroutputsrrrr4probsrriimg_himg_wrf scale_fctresultsslbscorelabelboxs r.post_process_object_detectionz1Owlv2ImageProcessor.post_process_object_detections( (:(:  #6{c,// j &b)u||,)/  #,- L%Aqad%AB L%Aqad%AB+2215 u99UE*D T4t$皙?g?r)rtarget_pred_boxesrHrMrIr&r8rrr rangeargsortrD unsqueezerrtensorrrrr zeros_likenonzeronumelrWr)rrrr nms_thresholdrr target_boxesrridxriousrrrralphas query_scores max_score query_alphasmask box_scoresr4s r.#post_process_image_guided_detectionz7Owlv2ImageProcessor.post_process_image_guided_detections, '~~w/H/H   #F s<7H(Hst t  # (:(:1(=(Bst t &b)u||,0 =  3 \//23 <s |4 "<#4QT#:#D#DQ#GVYIZ[\]^_`aD"DG8;F3K} 45 < <  #,- L%Aqad%AB L%Aqad%AB+2215 u UE5%$@aHKKLL_L_`I')AtQJ*??L!!&)++A./ SC!#;L'')//169L 1 2 ,/$6I(IO<SQL ::lC=L&F3K#;?DT*J %d+E NNjD5Q R' S*A&B%As J+/ J0)NN)TNNN)rN)rg333333?N)!__name__ __module__ __qualname____doc__model_input_namesrBILINEARboolrr,r)rstrrrrqrKarrayrr ndarrayrrFIRSTrrPILImagerrrr __classcell__)rts@r.r`r`s>(( ,3#'9'B'B!:>9=QQc5j)Q Q  Q 38n Q%QQU5$u+#567QE%e"456Q Q8?CDH xxeC)9$9:;$E#/?*?$@A F# >BDHAzzA38nA A eC)9$9:; A$E#/?*?$@AA AF%&# $!:>9=;?(8(>(>DHCCCCCC CC 38n CC  CCCCCCU5$u+#567CCE%e"456CC!sJ!78CC&CC$E#/?*?$@ACC CC'CCL_c9"'9=B:tTY{CZ=[9xQr0r`)5rrtypingrrrrrnumpyrKimage_processing_utilsr r r image_transformsr r r image_utilsrrrrrrrrrrrrutilsrrrrrr r!r&rr|r"r get_loggerrrr/r5rDrRr^r`ror0r.rs'55UU    $   H %G E  $N