sg#ddlmZmZddlmZmZmZmZmZm Z ddl m Z m Z er ddl mZddlmZerddlmZer dd lZdd lmZej,eZee d d Gd de Zy ))ListUnion)add_end_docstringsis_tf_availableis_torch_availableis_vision_availableloggingrequires_backends)Pipelinebuild_pipeline_init_args)Image) load_image)'TF_MODEL_FOR_VISION_2_SEQ_MAPPING_NAMESN)$MODEL_FOR_VISION_2_SEQ_MAPPING_NAMEST) has_tokenizerhas_image_processorcjeZdZdZfdZd dZd deeeededfffd Z d dZ dZ d Z xZ S) ImageToTextPipelinea Image To Text pipeline using a `AutoModelForVision2Seq`. This pipeline predicts a caption for a given image. Example: ```python >>> from transformers import pipeline >>> captioner = pipeline(model="ydshieh/vit-gpt2-coco-en") >>> captioner("https://huggingface.co/datasets/Narsil/image_dummy/raw/main/parrots.png") [{'generated_text': 'two birds are standing next to each other '}] ``` Learn more about the basics of using a pipeline in the [pipeline tutorial](../pipeline_tutorial) This image to text pipeline can currently be loaded from pipeline() using the following task identifier: "image-to-text". See the list of available models on [huggingface.co/models](https://huggingface.co/models?pipeline_tag=image-to-text). ct||i|t|d|j|jdk(r t yt y)Nvisiontf)super__init__r check_model_type frameworkrr)selfargskwargs __class__s W/var/www/html/venv/lib/python3.12/site-packages/transformers/pipelines/image_to_text.pyrzImageToTextPipeline.__init__EsD $)&)$) 7;~~7M 3 Sw ci}i}|||d<|||d<|||d<|"|d|vr td|j|||ifS)Nprompttimeoutmax_new_tokenszp`max_new_tokens` is defined both as an argument and inside `generate_kwargs` argument, please use only 1 version) ValueErrorupdate)rr'generate_kwargsr%r&forward_paramspreprocess_paramss r"_sanitize_parametersz(ImageToTextPipeline._sanitize_parametersLs  *0 h '  +2 i (  %/=N+ ,  &).>/.Q &  ! !/ 2 ."44r#inputsz Image.Imagec hd|vr|jd}| tdt| |fi|S)a Assign labels to the image(s) passed as inputs. Args: inputs (`str`, `List[str]`, `PIL.Image` or `List[PIL.Image]`): The pipeline handles three types of images: - A string containing a HTTP(s) link pointing to an image - A string containing a local path to an image - An image loaded in PIL directly The pipeline accepts either a single image or a batch of images. max_new_tokens (`int`, *optional*): The amount of maximum tokens to generate. By default it will use `generate` default. generate_kwargs (`Dict`, *optional*): Pass it to send all of these arguments directly to `generate` allowing full control of this function. timeout (`float`, *optional*, defaults to None): The maximum time in seconds to wait for fetching images from the web. If None, no timeout is set and the call may block forever. Return: A list or a list of list of `dict`: Each result comes as a dictionary with the following key: - **generated_text** (`str`) -- The generated text. imageszBCannot call the image-to-text pipeline without an inputs argument!)popr(r__call__)rr.r r!s r"r2zImageToTextPipeline.__call__asA< v ZZ)F >ab bw1&11r#ct||}|tjdt|tst dt |d|jjj}|dk(r|j||j}|jdk(r|j|j}|j|d j}|jj g|z}t#j$|j'd }|j)d |in|d k(rI|j|||j }|jdk(r|j|j}n|dk7rv|j||j}|jdk(r|j|j}|j||j}|j)|nVt d|d|j||j}|jdk(r|j|j}|jjjdk(r|d|d <|S)N)r&uPassing `prompt` to the `image-to-text` pipeline is deprecated and will be removed in version 4.48 of 🤗 Transformers. Use the `image-text-to-text` pipeline insteadz&Received an invalid text input, got - zy - but expected a single string. Note also that one single text can be provided for conditional image to text generation.git)r0return_tensorsptF)textadd_special_tokensr input_ids pix2struct)r0 header_textr5zvision-encoder-decoder)r5z Model type z- does not support conditional text generation)rlogger warning_once isinstancestrr(typemodelconfig model_typeimage_processorrto torch_dtype tokenizerr9 cls_token_idtorchtensor unsqueezer))rimager%r&rC model_inputsr9 text_inputss r" preprocesszImageToTextPipeline.preprocesss<5'2     W fc* >T)#/??43C3C#DL NN5NQ[[ !^^889IE !LL3==a@ ##[)$<=|+#335feieses3t >>T)#/??43C3C#DL77#335QUQ_Q_3` >>T)#/??43C3C#DL"nnVDNNnS ##K0!;zl:g!hii //uT^^/\L~~%+t/?/?@ ::   ' '5 0V^(,L %r#c d|vr-t|dtrtd|dDrd|d<d|vr|j|d<|j |j j }|j j|fi||}|S)Nr9c3$K|]}|du ywN).0xs r" z/ImageToTextPipeline._forward..sA!AIAsgeneration_config)r>listallrWr1rAmain_input_namegenerate)rrMr*r. model_outputss r"_forwardzImageToTextPipeline._forwards < '< 4d;A|K'@AA(,L % o 5373I3IO/ 0 !!$**"<"<=+ ++FVlVoV r#cxg}|D]2}d|jj|di}|j|4|S)Ngenerated_textT)skip_special_tokens)rGdecodeappend)rr\records output_idsrecords r" postprocesszImageToTextPipeline.postprocesssQ' #J $.."7"7(,#8#F NN6 " #r#)NNNNrR)NN)__name__ __module__ __qualname____doc__rr-rr?rr2rOr]rf __classcell__)r!s@r"rr-sF, 5*"2uS$s)]DDW%WX"2H1f, r#r)typingrrutilsrrrr r r baser rPILr image_utilsrmodels.auto.modeling_tf_autorrImodels.auto.modeling_autor get_loggerrgr<rrSr#r"rts{ 5(VP   H %,4UYZ[j(j\jr#