FPh\dZddlZddlZddlmcmZddlZddl m Z ddl m Z ddl mZddlmZmZddlmZdd lmZmZmZmZmZmZmZmZmZdd lmZGd d e Z y) a Generate predictions using the Segment Anything Model (SAM). SAM is an advanced image segmentation model offering features like promptable segmentation and zero-shot performance. This module contains the implementation of the prediction logic and auxiliary utilities required to perform segmentation using SAM. It forms an integral part of the Ultralytics framework and is designed for high-performance, real-time image segmentation tasks. N) LetterBox) BasePredictor)Results) DEFAULT_CFGops) select_device) batch_iteratorbatched_mask_to_boxbuild_all_layer_point_gridscalculate_stability_scoregenerate_crop_boxesis_box_near_crop_edgeremove_small_regionsuncrop_boxes_xyxy uncrop_masks) build_samceZdZdZeddffd ZdZdZddZddZ ddZ dd Z d Z fd Z d Zd ZdZeddZxZS) Predictora Predictor class for the Segment Anything Model (SAM), extending BasePredictor. The class provides an interface for model inference tailored to image segmentation tasks. With advanced architecture and promptable segmentation capabilities, it facilitates flexible and real-time mask generation. The class is capable of working with various types of prompts such as bounding boxes, points, and low-resolution masks. Attributes: cfg (dict): Configuration dictionary specifying model and task-related parameters. overrides (dict): Dictionary containing values that override the default configuration. _callbacks (dict): Dictionary of user-defined callback functions to augment behavior. args (namespace): Namespace to hold command-line arguments or other operational variables. im (torch.Tensor): Preprocessed input image tensor. features (torch.Tensor): Extracted image features used for inference. prompts (dict): Collection of various prompt types, such as bounding boxes and points. segment_all (bool): Flag to control whether to segment all objects in the image or only specified ones. Nc|i}|jtdddt| |||d|j_d|_d|_i|_d|_ y)a= Initialize the Predictor with configuration, overrides, and callbacks. The method sets up the Predictor object and applies any configuration overrides or callbacks provided. It initializes task-specific settings for SAM, such as retina_masks being set to True for optimal results. Args: cfg (dict): Configuration dictionary. overrides (dict, optional): Dictionary of values to override default configuration. _callbacks (dict, optional): Dictionary of callback functions to customize behavior. Nsegmentpredicti)taskmodeimgszTF) updatedictsuper__init__args retina_masksimfeaturesprompts segment_all)selfcfg overrides _callbacks __class__s iC:\Users\daisl\Desktop\realtime-object-detection\venv\Lib\site-packages\ultralytics/models/sam/predict.pyrzPredictor.__init__/sa  I99DIJ i4!%    c|j |jSt|tj }|rgt j |j |}|ddddfjd}t j|}tj|}|j|j}|jjr|jn|j}|r||j z |j"z }|S)a Preprocess the input image for model inference. The method prepares the input image by applying transformations and normalization. It supports both torch.Tensor and list of np.ndarray as input formats. Args: im (torch.Tensor | List[np.ndarray]): BCHW tensor format or list of HWC numpy arrays. Returns: (torch.Tensor): The preprocessed image tensor. N.)rr )r" isinstancetorchTensornpstack pre_transform transposeascontiguousarray from_numpytodevicemodelfp16halffloatmeanstd)r&r" not_tensors r+ preprocesszPredictor.preprocessEs 77 77N#B 55 $,,R01BC2I((6B%%b)B!!"%B UU4;; **//RWWYrxxz tyy.DHH,B r,ct|dk(sJdt|jjdd}|Dcgc] }|| c}Scc}w)a Perform initial transformations on the input image for preprocessing. The method applies transformations such as resizing to prepare the image for further preprocessing. Currently, batched inference is not supported; hence the list length should be 1. Args: im (List[np.ndarray]): List containing images in HWC numpy array format. Returns: (List[np.ndarray]): List of transformed images. r z6SAM model does not currently support batched inferenceF)autocenter)image)lenrr r)r&r" letterboxxs r+r6zPredictor.pre_transformasM2w!|UUU|diiooE%H ,./Bq "B///sAc,|jjd|}|jjd|}|jjd|}td|||fDr|j|g|i|S|j ||||||S)a Perform image segmentation inference based on the given input cues, using the currently loaded image. This method leverages SAM's (Segment Anything Model) architecture consisting of image encoder, prompt encoder, and mask decoder for real-time and promptable segmentation tasks. Args: im (torch.Tensor): The preprocessed input image in tensor format, with shape (N, C, H, W). bboxes (np.ndarray | List, optional): Bounding boxes with shape (N, 4), in XYXY format. points (np.ndarray | List, optional): Points indicating object locations with shape (N, 2), in pixel coordinates. labels (np.ndarray | List, optional): Labels for point prompts, shape (N, ). 1 for foreground and 0 for background. masks (np.ndarray, optional): Low-resolution masks from previous predictions. Shape should be (N, H, W). For SAM, H=W=256. multimask_output (bool, optional): Flag to return multiple masks. Helpful for ambiguous prompts. Defaults to False. Returns: (tuple): Contains the following three elements. - np.ndarray: The output masks in shape CxHxW, where C is the number of generated masks. - np.ndarray: An array of length C containing quality scores predicted by the model for each mask. - np.ndarray: Low-resolution logits of shape CxHxW for subsequent inference, where H=W=256. bboxespointsmasksc3$K|]}|du ywN.0is r+ z&Predictor.inference..s:"9QqDy"9s)r$popallgenerateprompt_inference) r&r"rLrMlabelsrNmultimask_outputr kwargss r+ inferencezPredictor.inferencers*!!(F3!!(F3   %0 :665"9: : 4==5d5f5 5$$RHXYYr,c|j|jj|n |j}|jddjdd|jdd} }|j rdnt | d|dz | d|dz } |tj|tj|j}|jdk(r|dn|}|"tj|jd}tj|tj|j}|| z}|dddddf|dddf}}|Ktj|tj|j}|jdk(r|dn|}|| z}|?tj|tj|jjd}|||fnd}|jj!|||\} } |jj#||jj j%| | |\} }| j'dd|j'ddfS) aJ Internal function for image segmentation inference based on cues like bounding boxes, points, and masks. Leverages SAM's specialized architecture for prompt-based, real-time segmentation. Args: im (torch.Tensor): The preprocessed input image in tensor format, with shape (N, C, H, W). bboxes (np.ndarray | List, optional): Bounding boxes with shape (N, 4), in XYXY format. points (np.ndarray | List, optional): Points indicating object locations with shape (N, 2), in pixel coordinates. labels (np.ndarray | List, optional): Labels for point prompts, shape (N, ). 1 for foreground and 0 for background. masks (np.ndarray, optional): Low-resolution masks from previous predictions. Shape should be (N, H, W). For SAM, H=W=256. multimask_output (bool, optional): Flag to return multiple masks. Helpful for ambiguous prompts. Defaults to False. Returns: (tuple): Contains the following three elements. - np.ndarray: The output masks in shape CxHxW, where C is the number of generated masks. - np.ndarray: An array of length C containing quality scores predicted by the model for each mask. - np.ndarray: Low-resolution logits of shape CxHxW for subsequent inference, where H=W=256. Nr rr0g?dtyper;)rMboxesrN)image_embeddingsimage_pesparse_prompt_embeddingsdense_prompt_embeddingsr[)r#r< image_encoderbatchshaper%minr2 as_tensorfloat32r;ndimr4onesint32 unsqueezeprompt_encoder mask_decoder get_dense_peflatten)r&r"rLrMrZrNr[r# src_shape dst_shapersparse_embeddingsdense_embeddings pred_masks pred_scoress r+rYzPredictor.prompt_inferences!&48==3H4::++B/dmm#zz!}Q/55bq9288AB<9 ##CYq\IaL-H)TU,YbcdYeJe)f  __V5==UF%+[[A%5VD\6F~a1__V5;;t{{SF aKF#AtQJ/4FF  __V5==UF%+[[A%5VD\6F aKF  OOEt{{S]]^_`E%+%7&&!T.2jj.G.Gv]ckp.G.q++#'**"9"9%ZZ..;;=%6$4- #:#  K!!!Q')<)@PQ (,(=(=gfgk(=(l% :MM)D/Aq6 bghijk  :-(1# 3: ";ItzzG`G`>(1# 3: % (A(AA / :@@B 29kAqRTVX>ZZ yy+7@7KYW`Macmnwcx*yI!!),""9-"":.+R0:.J))K0K))K0K??&&{KOD+K,={KK%j&6 RLJ%d+K   j )   { +   { +    C O < =['D^YYz* ii , ii , yy.  | q %F??&&{FOLD3=d3C[QUEVXcdhXi[ J; 33r,cht|jj|}|t|jj}|j |j ||_||_tjgdjdddj ||_ tjgdjdddj ||_ d|j_ d|j_ d|j_d|j_d |_y) a[ Initializes the Segment Anything Model (SAM) for inference. This method sets up the SAM model by allocating it to the appropriate device and initializing the necessary parameters for image normalization and other Ultralytics compatibility settings. Args: model (torch.nn.Module): A pre-trained SAM model. If None, a model will be built based on configuration. verbose (bool): If True, prints selected device information. Attributes: model (torch.nn.Module): The SAM model allocated to the chosen device for inference. device (torch.device): The device to which the model and tensors are allocated. mean (torch.Tensor): The mean values for image normalization. std (torch.Tensor): The standard deviation values for image normalization. )verboseN)g33333^@gR]@gRY@r.r )g(\2M@g(\L@gL@F T)rr r;rr<evalr:r2rviewr@rApttritonstrider= done_warmup)r&r<rr;s r+ setup_modelzPredictor.setup_model1s"tyy//A =diioo.E XXf%  LL!:;@@QJMMfU << 78==b!QGJJ6R !   r,c p|dd\}}|jr|dnd}ttdtt |D}t |t stj|}g}t|gD]/\} } || } |tj|jdd|j| jd}tjt |tj|j} tj ||dddf| dddfgd}tj"| dj| jdddd } | |j$j&kD} |j(d | } |j+t-| | || | 2d|_|S) a Post-processes SAM's inference outputs to generate object detection masks and bounding boxes. The method scales masks and boxes to the original image size and applies a threshold to the mask predictions. The SAM model uses advanced architecture and promptable segmentation tasks to achieve real-time performance. Args: preds (tuple): The output from SAM model inference, containing masks, scores, and optional bounding boxes. img (torch.Tensor): The processed input image tensor. orig_imgs (list | torch.Tensor): The original, unprocessed images. Returns: (list): List of Results objects containing detection masks, bounding boxes, and other metadata. Nr0c32K|]}t|ywrP)strrRs r+rUz(Predictor.postprocess..dsF/E!s1v/EsF)paddingr_r.dimr)pathnamesrNra)r%r enumeraterangerHr1listrconvert_torch2numpy_batch scale_boxesrhr?r2arangernr;r scale_masksr<rrgrr)r&predsimg orig_imgsryrzrrresultsrTrNorig_imgclsimg_paths r+ postprocesszPredictor.postprocessRs #() K"&"2"2eAh YFuS_/EFFG)T*55i@I!:,/HAu |H&!oociim[=N=N=PRZR`R`jop ll3z?%++jN_N_`#iik!T'6JCPQSWPWL(Y_ab OOE$K$5$5$79KUZ[\]^EDJJ555Ezz!}Q'H NN78(%u\gh i0!r,c*|t||yy)aW Sets up the data source for inference. This method configures the data source from which images will be fetched for inference. The source could be a directory, a video file, or other types of image data sources. Args: source (str | Path): The path to the image data source for inference. N)r setup_source)r&sourcer*s r+rzPredictor.setup_sourceys   G  ( r,cv|j0t|jj}|j||j |t |j dk(sJd|j D]>}|j|d}|jj||_ ||_ yy)a Preprocesses and sets a single image for inference. This function sets up the model if not already initialized, configures the data source to the specified image, and preprocesses the image for feature extraction. Only one image can be set at a time. Args: image (str | np.ndarray): Image file path as a string, or a np.ndarray image read by cv2. Raises: AssertionError: If more than one image is set. Nr z,`set_image` only supports setting one image!) r<rr rrrHdatasetrCrfr#r")r&rGr<rgr"s r+ set_imagezPredictor.set_images :: diioo.E   U # % 4<< A%U'UU%\\Eq*B JJ44R8DMDG  "r,c||_y)zSet prompts in advance.N)r$)r&r$s r+ set_promptszPredictor.set_promptss  r,c d|_d|_y)z*Resets the image and its features to None.N)r"r#)r&s r+ reset_imagezPredictor.reset_images r,ct|dk(r|Sg}g}|D]}|jjjtj }t ||d\}}| }t ||d\}}|xr| }|jtj|jd|jt|tj|d}t|}tjj!|jtj||} || j#|j$|j&| fS)a Perform post-processing on segmentation masks generated by the Segment Anything Model (SAM). Specifically, this function removes small disconnected regions and holes from the input masks, and then performs Non-Maximum Suppression (NMS) to eliminate any newly created duplicate boxes. Args: masks (torch.Tensor): A tensor containing the masks to be processed. Shape should be (N, H, W), where N is the number of masks, H is height, and W is width. min_area (int): The minimum area below which disconnected regions and holes will be removed. Defaults to 0. nms_thresh (float): The IoU threshold for the NMS algorithm. Defaults to 0.7. Returns: (tuple([torch.Tensor, List[int]])): - new_masks (torch.Tensor): The processed masks with small regions removed. Shape is (N, H, W). - keep (List[int]): The indices of the remaining masks post-NMS, which can be used to filter the boxes. rholes)rislandsr)r;r`)rHcpunumpyastyper4uint8rrr2rjror?rr rrrr:r;r`) rNmin_area nms_thresh new_masksrmaskchanged unchangedrars r+rzPredictor.remove_small_regionss!$ u:?L D88:##%,,RXX6D0xgNMD'# I0xiPMD'!1'kI   U__T2<rsO .6..7uuutP tPr,