qi,&dZddlmZddlmZddlmZmZddlm Z m Z ddl m Z m Z mZmZmZmZddlmZdd lmZmZgd Zgd ZGd d edZdZGddeZdgZy)z"Image processor class for Idefics.)Callable)Image)BaseImageProcessor BatchFeature)resizeto_channel_dimension_format)ChannelDimension ImageInputPILImageResamplingmake_flat_list_of_imagesto_numpy_array valid_images) ImagesKwargs) TensorTypeis_torch_available)g3<4'?gwg M?gy{ ?)gB91?gwt.?g U?c6eZdZUdZedzed<eeefed<y)IdeficsImageProcessorKwargsa transform (`Callable`, *optional*): A custom transform function that accepts a single image can be passed for training. For example, `torchvision.Compose` can be used to compose multiple transforms. If `None` - an inference mode is assumed - and then a preset of inference-specific transforms will be applied to the images image_size (`dict[str, int]`, *optional*): Resize to image size N transform image_size) __name__ __module__ __qualname____doc__r__annotations__dictstrintf/opt/pipecat/venv/lib/python3.12/site-packages/transformers/models/idefics/image_processing_idefics.pyrr&s!$S#Xr rF)totalc|jdk(r|S|jd}tjd|jd}tj ||}|jd}|S)NRGBRGBA)r&r&)modeconvertrnewsizealpha_composite)image image_rgba backgroundr+s r!convert_to_rgbr/4sa zzU v&J6:??ODJ++J CO%--e4O r c6eZdZdZdgZeZ ddedee ezdzdee ezdzdedzd e d eezd dffd Z ddddddde jfd ededzdeeefdzdee ezdzdee ezdzdedzd e dzd edzdee zdzd e fdZxZS)IdeficsImageProcessora Constructs a Idefics image processor. Args: image_size (`int`, *optional*, defaults to 224): Resize to image size image_mean (`float` or `list[float]`, *optional*, defaults to `IDEFICS_STANDARD_MEAN`): Mean to use if normalizing the image. This is a float or list of floats the length of the number of channels in the image. Can be overridden by the `image_mean` parameter in the `preprocess` method. Can be overridden by the `image_mean` parameter in the `preprocess` method. image_std (`float` or `list[float]`, *optional*, defaults to `IDEFICS_STANDARD_STD`): Standard deviation to use if normalizing the image. This is a float or list of floats the length of the number of channels in the image. Can be overridden by the `image_std` parameter in the `preprocess` method. Can be overridden by the `image_std` parameter in the `preprocess` method. image_num_channels (`int`, *optional*, defaults to 3): Number of image channels. do_rescale (`bool`, *optional*, defaults to `True`): Whether to rescale the image by the specified scale `rescale_factor`. Can be overridden by `do_rescale` in the `preprocess` method. rescale_factor (`int` or `float`, *optional*, defaults to `1/255`): Scale factor to use if rescaling the image. Can be overridden by `rescale_factor` in the `preprocess` method. pixel_valuesNrr image_mean image_stdimage_num_channels do_rescalerescale_factorreturnc t|di|||_||_||nt|_||nt |_||_||_ y)Nr) super__init__rr5IDEFICS_STANDARD_MEANr3IDEFICS_STANDARD_STDr4r6r7) selfrr3r4r5r6r7kwargs __class__s r!r;zIdeficsImageProcessor.__init__]sQ "6"$"4(2(>*DY&/&;AU$,r imagesrreturn_tensorsc  ||n |j}||n |j}||n |j}||n |j}||n |j}||n |j }||f} t |trt|dk(rgS|j|}t|}t|s td|?ts tdddl} |D cgc] } ||  }} | j!|S|D cgc] } t#| }} |D cgc] } t%| }} |D cgc]} t'| | t(j* }} |Dcgc]}|j-||}}|D cgc]} |j/| ||}} |D cgc]} t1| t2j4}} t7d|i| d}|Scc} wcc} wcc} wcc} wcc}wcc} wcc} w) a| Preprocess a batch of images. Args: images (`ImageInput`): A list of images to preprocess. image_size (`int`, *optional*, defaults to `self.image_size`): Resize to image size image_num_channels (`int`, *optional*, defaults to `self.image_num_channels`): Number of image channels. image_mean (`float` or `list[float]`, *optional*, defaults to `IDEFICS_STANDARD_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. Can be overridden by the `image_mean` parameter in the `preprocess` method. image_std (`float` or `list[float]`, *optional*, defaults to `IDEFICS_STANDARD_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. Can be overridden by the `image_std` parameter in the `preprocess` method. transform (`Callable`, *optional*, defaults to `None`): A custom transform function that accepts a single image can be passed for training. For example, `torchvision.Compose` can be used to compose multiple transforms. If `None` - an inference mode is assumed - and then a preset of inference-specific transforms will be applied to the images do_rescale (`bool`, *optional*, defaults to `True`): Whether to rescale the image by the specified scale `rescale_factor`. Can be overridden 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 overridden by `rescale_factor` in the `preprocess` method. Returns: a PyTorch tensor of the processed images NrzSInvalid image type. Must be of type PIL.Image.Image, numpy.ndarray, or torch.Tensorz.To pass in `transform` torch must be installed)resample)r,scale)meanstdr2)data tensor_type)rr5r3r4r6r7 isinstancelistlen fetch_imagesr r ValueErrorr ImportErrortorchstackr/rrr BICUBICrescale normalizer r FIRSTr)r>rAr5rr3r4rr6r7rBr?r*rPxr,s r! preprocessz IdeficsImageProcessor.preprocessps\$.#9Zt 3E3Q/W[WnWn#-#9Zt !*!6IDNN #-#9Zt +9+E4K^K^J' fd #F q(8I""6*)&1F#rs s  %'!"RSS ,23qil3F3;;v& &.44.#44-34.#44PVW1&D+=+E+EFWWOUVe$,,U.,AVVMST$...CTTRXYQ-a1A1G1GHYYNF#;XYgh 454WVTYs*G <GG,#GG!5G&!G+)NNrTgp?)rrrrmodel_input_namesr valid_kwargsrfloatrKboolr;rPYTORCHr rrrrW __classcell__)r@s@r!r1r1Asn0((.L1504)*&---DK'$.-4;&- -  $J -  -e - -,*+,01504%)"&'+2<2D2DYY $JYcNT) Y DK'$. Y 4;&- Yd?Y4KY Yj(4/Y Yr r1N)rcollections.abcrPILrimage_processing_utilsrrimage_transformsrr image_utilsr r r r rrprocessing_utilsrutilsrrr<r=rr/r1__all__rr r!rgsc)$FC-3<; ,e  H.HV # #r