U !?h{,@sddlZddlmZddlmZddlZddlZddlZ ddl Z ddl m Z ddl mZddlmZddlmZeeedd d Zejd d d ZddZddZGdddeZdS)N)deepcopy)List)validate)Image) BaseSession)oldholdwlong_side_lengthcCsD|dt||}||||}}t|d}t|d}||fS)Ng?g?)maxint)rr r scaleZnewhZnewwrD/var/www/html/venv/lib/python3.8/site-packages/rembg/sessions/sam.pyget_preprocess_shapes   r)coordscCsZ|\}}t|d|d|\}}t|t}|d|||d<|d|||d<|S)Nrr).r).r)rrastypefloat)r original_sizeZ target_lengthZold_hZold_wZnew_hZnew_wrrr apply_coordssrcCsg}g}|D]}|ddkr:||d||dq |ddkr ||dd|ddg||dd|dd g|d|d q t|t|}}||fS) NtypepointdatalabelZ rectanglerr)appendnparray)promptZpointslabelsmarkrrrget_input_points&s    r"cCsg}t|jdD]d}g}t|jdD]B}|||f}tj||dd|d|dftjd}||q(||qt|S)Nrrrflags)rangeshapecv2 warpAffine INTER_LINEARrrr)masksrtransform_matrixZ output_masksbatchZ batch_masksZmask_idmaskrrrtransform_masks8s    r.c@sPeZdZdZd eejdddZee edddZ e d d Z e d d Z dS) SamSessiona This class represents a session for the Sam model. Args: model_name (str): The name of the model. sess_opts (ort.SessionOptions): The session options. *args: Variable length argument list. **kwargs: Arbitrary keyword arguments. N) model_name sess_optsc Os||_g}t}|r8|pgD]}||kr||qn |||jj||} tjt| d||d|_ tjt| d||d|_ dS)aF Initialize a new SamSession with the given model name and session options. Args: model_name (str): The name of the model. sess_opts (ort.SessionOptions): The session options. *args: Variable length argument list. **kwargs: Arbitrary keyword arguments. r) providersZ sess_optionsrN) r0ortZget_available_providersrextend __class__download_modelsZInferenceSessionstrencoderdecoder) selfr0r1r2argskwargsZvalid_providersZavailable_providersproviderpathsrrr__init__Vs&    zSamSession.__init__)imgreturnc Os|dddt|jdt|jdgdg}dddd idd iddd id dd d }t||dd}d}|jdj}|d}t |} | j dd} |d| j d} |d| j d} t | | } t | ddgd| dgdddgg}t j| |dd|d|dft jd} || t ji}|jd|}|d}|| |d}t|\}}t j|t ddgggdddddddf}t j|t dggdddddft j}t|||t j}t j|t jd|j ddft jdgdd}t ||j}|ddddddft j}t jdt jd}t jdt jd}|||||t j |t jdd}|jd|\}}}t j|}t|| |}t j|j d|j ddft jd}|dddddddfD]}dddg||dk<qt |dgS)a} Predict masks for an input image. This function takes an image as input and performs various preprocessing steps on the image. It then runs the image through an encoder to obtain an image embedding. The function also takes input labels and points as additional arguments. It concatenates the input points and labels with padding and transforms them. It creates an empty mask input and an indicator for no mask. The function then passes the image embedding, point coordinates, point labels, mask input, and has mask input to a decoder. The decoder generates masks based on the input and returns them as a list of images. Parameters: img (PILImage): The input image. *args: Additional arguments. **kwargs: Additional keyword arguments. Returns: List[PILImage]: A list of masks generated by the decoder. Z sam_promptrrr)rrrrobjectrstringintegernumber)ritems)r properties)instanceschema)irJrRGBNr#)image_embeddingrr+g)Zaxis)Zdtype)rrrN)Zimage_embeddingsZ point_coordsZ point_labelsZ mask_inputZhas_mask_inputZ orig_im_sizerL)!getr widthheightrr8 get_inputsnameconvertrrr&minr'r(r)rZfloat32runr"Z concatenaterZonesmatmulTZzerosr9Zlinalginvr.Zuint8rZ fromarray)r:r@r;r<rrIZ target_sizeZ input_sizeZencoder_input_nameZcv_imagerZscale_xZscale_yr r+Zencoder_inputsZencoder_outputrLZ embeddingZ input_pointsZ input_labelsZ onnx_coordZ onnx_labelZonnx_mask_inputZonnx_has_mask_inputZdecoder_inputsr*_Zinv_transform_matrixr-mrrrpredicts           "   " zSamSession.predictc Os|dd}|dd}|d}|d}|rD|d}|d}tjd |d ||j||d d tjd |d ||j||d d |d krbtjtj|j||dsbt}t ddD]l}tjd|dd d|d|j||d d tj|j||d|d} | t | d t | qt tj|j||dd} | |W5QRXtj|j|||tj|j|||fS)ad Class method to download ONNX model files. This method is responsible for downloading two ONNX model files from specified URLs and saving them locally. The downloaded files are saved with the naming convention 'name_encoder.onnx' and 'name_decoder.onnx', where 'name' is the value returned by the 'name' method. Parameters: cls: The class object. *args: Variable length argument list. **kwargs: Arbitrary keyword arguments. Returns: tuple: A tuple containing the file paths of the downloaded encoder and decoder models. Z sam_modelZsam_vit_b_01ec64Z sam_quantFz .encoder.onnxz .decoder.onnxz.encoder.quant.onnxz.decoder.quant.onnxz>https://github.com/danielgatis/rembg/releases/download/v0.0.0/NT)fnamepathZ progressbarzsam_vit_h_4b8939.encoder.onnxz!sam_vit_h_4b8939.encoder_data.binrz\https://github.com/danielgatis/rembg/releases/download/v0.0.0/sam_vit_h_4b8939.encoder_data.z.binzsam_vit_h_4b8939.encoder_data.rbwb)rQpoochretrieveZ u2net_homeosr`existsjoin bytearrayr%r4openreadremovewrite) clsr;r<r0ZquantZ fname_encoderZ fname_decodercontentiZfbinfprrrr6sl                zSamSession.download_modelscOsdS)aE Class method to return a string value. This method returns the string value 'sam'. Parameters: cls: The class object. *args: Variable length argument list. **kwargs: Arbitrary keyword arguments. Returns: str: The string value 'sam'. Zsamr)rnr;r<rrrrURszSamSession.name)N)__name__ __module__ __qualname____doc__r7r3ZSessionOptionsr?PILImagerr^ classmethodr6rUrrrrr/Ks +  Mr/)rfcopyrtypingrr'numpyrZ onnxruntimer3rdZ jsonschemarZPILrZ PIL.Imagervbaserr rZndarrayrr"r.r/rrrrs