tj2^ddlZddlmZddlmZmZddlmZdZGdd eZ d gZ dS) N)'SemanticSegmentationPostProcessorOutput)IMAGENET_STANDARD_MEANIMAGENET_STANDARD_STD)Sam2ImageProcessorct|ttfr=tjd|D}tjd|D}nBt|tjr|d\}}ntdtj||||gd}| d |j }||z}|S)a Scale batch of bounding boxes to the target sizes. Args: boxes (`torch.Tensor` of shape `(batch_size, num_boxes, 4)`): Bounding boxes to scale. Each box is expected to be in (x1, y1, x2, y2) format. target_sizes (`list[tuple[int, int]]` or `torch.Tensor` of shape `(batch_size, 2)`): Target sizes to scale the boxes to. Each target size is expected to be in (height, width) format. Returns: `torch.Tensor` of shape `(batch_size, num_boxes, 4)`: Scaled bounding boxes. cg|] }|d S)r.0is c/var/www/html/CA-Chatbot/venv/lib/python3.11/site-packages/transformers/models/sam3/modular_sam3.py z _scale_boxes..)s$@$@$@aQqT$@$@$@cg|] }|d S)r r s rrz _scale_boxes..*s#?#?#?QAaD#?#?#?rrz4`target_sizes` must be a list, tuple or torch.Tensordim) isinstancelisttupletorchtensorTensorunbind TypeErrorstack unsqueezetodevice)boxes target_sizes image_height image_width scale_factors r _scale_boxesr's,u ..P|$@$@<$@$@$@AA l#?#?,#?#?#?@@ L%, / /P$0$7$7$:$:! kkNOOO; \; U[\]]]L))!,,// ==L L E Lrc eZdZeZeZdddZdddZ dde e e e fdzde d e d d fd Zdde de e dzfdZ dde de de e dzfdZdS)Sam3ImageProcessori)heightwidthi N?Fr# thresholdreturn_segmentation_scoresreturnzBlist[torch.Tensor] | list[SemanticSegmentationPostProcessorOutput]c |j}|td| t|}|t|t|krtdg}t |D]}t jj | d||dd} | t| d k t j | dd  n fd t |D}|s d |D}|S)a Converts the output of [`Sam3Model`] into semantic segmentation maps. Args: outputs ([`Sam3ImageSegmentationOutput`]): Raw outputs of the model containing semantic_seg. target_sizes (`list[tuple[int, int]]` of length `batch_size`, *optional*): List of tuples corresponding to the requested final size (height, width) of each prediction. If unset, predictions will not be resized. threshold (`float`, *optional*, defaults to 0.5): Threshold for binarizing the semantic segmentation masks. return_segmentation_scores (`bool`, *optional*, defaults to `False`): Whether to return segmentation scores alongside the segmentation map. When `True`, each element of the returned list is a [`SemanticSegmentationPostProcessorOutput`] with fields `segmentation` (binary class IDs, shape `(height, width)`) and `segmentation_scores` (sigmoid probabilities, shape `(1, height, width)`). Returns: `list[torch.Tensor]` or `list[SemanticSegmentationPostProcessorOutput]`: When `return_segmentation_scores=False` (default), a list of length `batch_size` where each item is a segmentation map of shape `(height, width)` with class IDs. When `return_segmentation_scores=True`, a list of [`SemanticSegmentationPostProcessorOutput`] with fields `segmentation` (class IDs, shape `(height, width)`) and `segmentation_scores` (shape `(1, height, width)`). In both cases, `(height, width)` corresponds to the target size (if `target_sizes` is specified). NzSemantic segmentation output is not available in the model outputs. Make sure the model was run with semantic segmentation enabled.zTMake sure that you pass in as many target sizes as the batch dimension of the logitsrrbilinearFsizemode align_corners)rr segmentationsegmentation_scoresdatacg|]D}t|dfktj|dES)rr6r9)rr rlong)r rsemantic_probsr-s rrzISam3ImageProcessor.post_process_semantic_segmentation..sm%%% 8)71)= )I(M(Mej(Y(Y/=a/@%%%rcg|] }|j Sr )r7)r items rrzISam3ImageProcessor.post_process_semantic_segmentation..s$Y$Y$Y4T%6$Y$Y$Yr) semantic_seg ValueErrorsigmoidlenrangernn functional interpolaterappendrr r<) selfoutputsr#r-r.semantic_logits batch_sizesemantic_segmentationidx resized_probsr=s ` @r"post_process_semantic_segmentationz5Sam3ImageProcessor.post_process_semantic_segmentation<sD".  "R  )0022))   #?##s<'8'888 j%' !Z((   % 3 ? ?"3'11a188%c*#"' !@!! &,,;-:4-@9-L,P,PQVQ[,\,\3@3C  %%%%%z**%%% !* Z$Y$YCX$Y$Y$Y !$$r333333?c|j}|j}|j}|jd}|"t ||krt d|}||} || z}|} |t| |} g} t|| D]3\} } | |k}| |} | |} | | | d4| S)aD Converts the raw output of [`Sam3Model`] into final bounding boxes in (top_left_x, top_left_y, bottom_right_x, bottom_right_y) format. Args: outputs ([`Sam3ImageSegmentationOutput`]): Raw outputs of the model containing pred_boxes, pred_logits, and optionally presence_logits. threshold (`float`, *optional*, defaults to 0.3): Score threshold to keep object detection predictions. target_sizes (`list[tuple[int, int]]`, *optional*): 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 following keys: - **scores** (`torch.Tensor`): The confidence scores for each predicted box on the image. - **boxes** (`torch.Tensor`): Image bounding boxes in (top_left_x, top_left_y, bottom_right_x, bottom_right_y) format. rN9Make sure that you pass in as many target sizes as images)scoresr") pred_logits pred_boxespresence_logitsshaperCrArBr'ziprH)rIrJr-r#rUrVrWrL batch_scorespresence_scores batch_boxesresultsrTr"keeps rpost_process_object_detectionz0Sam3ImageProcessor.post_process_object_detections() ' !1 &q)  #L(9(9Z(G(GXYY Y#**,,  &-5577O'/9L!   #&{LAAK {;; ? ?MFEI%DD\F$KE NNfu== > > > >rmask_thresholdc,|j}|j}|j}|j}|jd} |"t || krt d|} ||} | | z} |} |} |t| |} g}tt| | | D]\}\}}}||k}||}||}||}|i||}t |dkrNtj j |d|ddd}||ktj}||||d|S)aQ Converts the raw output of [`Sam3Model`] into instance segmentation predictions with bounding boxes and masks. Args: outputs ([`Sam3ImageSegmentationOutput`]): Raw outputs of the model containing pred_boxes, pred_logits, pred_masks, and optionally presence_logits. threshold (`float`, *optional*, defaults to 0.3): Score threshold to keep instance predictions. mask_threshold (`float`, *optional*, defaults to 0.5): Threshold for binarizing the predicted masks. target_sizes (`list[tuple[int, int]]`, *optional*): List of tuples (`tuple[int, int]`) containing the target size `(height, width)` of each image in the batch. If unset, predictions will not be resized. Returns: `list[dict]`: A list of dictionaries, each dictionary containing the following keys: - **scores** (`torch.Tensor`): The confidence scores for each predicted instance on the image. - **boxes** (`torch.Tensor`): Image bounding boxes in (top_left_x, top_left_y, bottom_right_x, bottom_right_y) format. - **masks** (`torch.Tensor`): Binary segmentation masks for each instance, shape (num_instances, height, width). rNrSr1Fr2)rTr"masks)rUrV pred_masksrWrXrCrArBr' enumeraterYrrErFrGrsqueezer r<rH)rIrJr-r`r#rUrVrcrWrLrZr[ batch_masksr\r]rNrTr"rbr^ target_sizes r"post_process_instance_segmentationz5Sam3ImageProcessor.post_process_instance_segmentations<) ' ' !1 &q)  #L(9(9Z(G(GXYY Y#**,,  &-5577O'/9L!((** !   #&{LAAK+4S{T_5`5`+a+a O O 'C'&%I%DD\F$KE$KE'*3/ u::>>!H/;;**('&+ < gajj ^+// ;;E NNfuuMM N N N Nr)Nr,F)rQN)rQr,N)__name__ __module__ __qualname__r image_meanr image_stdr3 mask_sizerrintfloatboolrPr_rhr rrr)r)6s+'J%IT * *D--I 6:+0 S%S%5c?+d2S% S% %) S% N S%S%S%S%j111[_`e[fim[m1111l #+/ PPP P 5kD( PPPPPPrr)) rimage_processing_outputsr image_utilsrrsam2.image_processing_sam2rr'r)__all__r rrrvs  OOOOOO<;;;;;8^^^^^+^^^B r