tj3 ,ddlmZddlZddlZddlmZddlmZddl m Z ddl m Z ddl mZdd lmZmZmZdd lmZmZmZejeZ dd eeefd ededzdedeeeff dZGdde ZGdde ZddgZdS))OptionalN)nn) functional)DetrImageProcessor)DetrImageProcessorPil)center_to_corners_format)PILImageResamplingSizeDict#get_image_size_for_max_height_width) TensorTypeloggingrequires_backends image_sizesizemax_sizemod_sizereturnc$|\}}d}|ltt||f}tt||f}||z |z|kr$||z|z }tt |}||kr2|} ||t||z|z } ndt||z|z } nN||kr||ks ||kr ||kr||} } n1|} ||t||z|z } nt||z|z } || | |zz } | | |zz } | | fS)as Computes the output image size given the input image size and the desired output size, while ensuring that both height and width are multiples of `mod_size`. This mirrors the YOLOS-specific behavior used in the torch/fast backends and is required so that all YOLOS image processing backends (PIL, torchvision, fast) produce identical output shapes. N)floatminmaxintround) rrrrheightwidthraw_sizemin_original_sizemax_original_sizeowohs e/var/www/html/CA-Chatbot/venv/lib/python3.11/site-packages/transformers/models/yolos/modular_yolos.py get_size_with_aspect_ratio_yolosr$sfMFEH!#vuo"6"677!#vuo"6"677 0 04 7( B B"336GGHuX''D v~~   H$8X&.//BBTF]U*++BB E//fnn%6//etmmB   H$8X%.//BBTE\F*++B 2= ! 2= ! 8Oc eZdZ ddejdededdejffd Z dd ed e e e zfd Z d Z d ZdZxZS)YolosImageProcessorNimagerresamplez0PILImageResampling | tvF.InterpolationMode | intrc |jr0|jr)t|jdd|j|j}ng|jr0|jr)t |jdd|j|j}n0|jr|jr|j|jf}ntd|dtj |ft|d|d|d|}|S) a[ Resize the image to the given size. Size can be `min_size` (scalar) or `(height, width)` tuple. If size is an int, smaller edge of the image will be matched to this number. Args: image (`torch.Tensor`): Image to resize. size (`SizeDict`): Size of the image's `(height, width)` dimensions after resizing. Available options are: - `{"height": int, "width": int}`: The image will be resized to the exact size `(height, width)`. Do NOT keep the aspect ratio. - `{"shortest_edge": int, "longest_edge": int}`: The image will be resized to a maximum size respecting the aspect ratio and keeping the shortest edge less or equal to `shortest_edge` and the longest edge less or equal to `longest_edge`. - `{"max_height": int, "max_width": int}`: The image will be resized to the maximum size respecting the aspect ratio and keeping the height less or equal to `max_height` and the width less or equal to `max_width`. resample (`PILImageResampling | tvF.InterpolationMode | int`, *optional*, defaults to `PILImageResampling.BILINEAR`): Resampling filter to use if resizing the image. N\Size must contain 'height' and 'width' keys or 'shortest_edge' and 'longest_edge' keys. Got .rrrrr)) shortest_edge longest_edger$shape max_height max_widthr rr ValueErrorsuperresizer selfr(rr)kwargsnew_size __class__s r#r8zYolosImageProcessor.resize=s 6   $"3 8 BCC8H$J\^b^oppHH _  :5;rss;KT_^b^lmmHH [ TZ  TZ0HHvosvvv     8A;GGGRZ  ^d   r%? threshold target_sizesc4|j|j}}|/t|t|krtdtj|d}|dddfd\}}t|} |t|tr=tj d|D} tj d|D} n| d\} } tj| | | | gd| j} | | dddddfz} g} t#||| D]C\}}}|||k}|||k}|||k}| |||d D| S)  Converts the raw output of [`YolosForObjectDetection`] into final bounding boxes in (top_left_x, top_left_y, bottom_right_x, bottom_right_y) format. Only supports PyTorch. Args: outputs ([`YolosObjectDetectionOutput`]): Raw outputs of the model. threshold (`float`, *optional*): Score threshold to keep object detection predictions. target_sizes (`torch.Tensor` or `list[tuple[int, int]]`, *optional*): Tensor of shape `(batch_size, 2)` or 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 scores, labels and boxes for an image in the batch as predicted by the model. NTMake sure that you pass in as many target sizes as the batch dimension of the logits.cg|] }|d Sr.0is r# zEYolosImageProcessor.post_process_object_detection..%A%A%Aqad%A%A%Ar%cg|] }|d Sr.rGrHs r#rKzEYolosImageProcessor.post_process_object_detection..rLr%r.dimscoreslabelsboxes)logits pred_boxeslenr6rrsoftmaxrr isinstancelisttorchTensorunbindstacktodevicezipappendr:outputsr?r@ out_logitsout_bboxprobrRrSrTimg_himg_w scale_fctresultsslbscorelabelboxs r#post_process_object_detectionz1YolosImageProcessor.post_process_object_detectionjs& '~w/AH  #:#l"3"333 j}$$Z44c3B3h++B//)22  #,-- 6 %A%AL%A%A%ABB %A%AL%A%A%ABB+22155 u UE5%$@aHHHKKELYYIIaaaqqqj11E66511 M MGAq!a)m$Ea)m$EA M"C NNeusKK L L L Lr%c tdNzHSegmentation post-processing is not implemented for Deformable DETR yet.NotImplementedErrorr:s r#"post_process_instance_segmentationz6YolosImageProcessor.post_process_instance_segmentation!"lmmmr%c tdNzQSemantic segmentation post-processing is not implemented for Deformable DETR yet.rurws r#"post_process_semantic_segmentationz6YolosImageProcessor.post_process_semantic_segmentation!"uvvvr%c tdNzQPanoptic segmentation post-processing is not implemented for Deformable DETR yet.rurws r#"post_process_panoptic_segmentationz6YolosImageProcessor.post_process_panoptic_segmentationr}r%Nr>N)__name__ __module__ __qualname__r[r\r rr8rr rZtuplerrrxr|r __classcell__r=s@r#r'r'<s RV ++|++MN +  ++++++\Y]33"'3=G$u+=U3333jnnnwwwwwwwwwwr%r'c eZdZ ddejdededdejffd Z dd ed e e e zfd Z d Z d ZdZxZS)YolosImageProcessorPilNr(rr)r rc ||n|j}|jr7|jr0t|jdd|j|jp|j}ng|jr0|jr)t|jdd|j|j}n0|jr|j r|j|j f}ntd|dtj |ft|d|d|d|}|S) a; Resize the image to the given size. Size can be `min_size` (scalar) or `(height, width)` tuple. If size is an int, smaller edge of the image will be matched to this number. Args: image (`np.ndarray`): Image to resize. size (`SizeDict`): Size of the image's `(height, width)` dimensions after resizing. Available options are: - `{"height": int, "width": int}`: The image will be resized to the exact size `(height, width)`. Do NOT keep the aspect ratio. - `{"shortest_edge": int, "longest_edge": int}`: The image will be resized to a maximum size respecting the aspect ratio and keeping the shortest edge less or equal to `shortest_edge` and the longest edge less or equal to `longest_edge`. - `{"max_height": int, "max_width": int}`: The image will be resized to the maximum size respecting the aspect ratio and keeping the height less or equal to `max_height` and the width less or equal to `max_width`. resample (`PILImageResampling`, *optional*, defaults to `PILImageResampling.BILINEAR`): Resampling filter to use if resizing the image. Nr+r,r-rr.r/r0)r)r1r2r$r3r4r5r rrr6r7r8r r9s r#r8zYolosImageProcessorPil.resizes.6 (388   $"3 8 BCC "!7T%7HH _  :5;rss;KT_^b^lmmHH [ TZ  TZ0HHvosvvv   !HQK@@@       r%r>r?r@cVt|dg|j|j}}|/t|t|krt dt j|d}|dddfd\}}t|} |t|tr=tj d|D} tj d|D} n|d\} } tj| | | | gd | j} | | dddddfz} g} t%||| D]C\}}}|||k}|||k}|||k}| |||d D| S) rBr[NrCrD.cg|] }|d SrFrGrHs r#rKzHYolosImageProcessorPil.post_process_object_detection..rLr%cg|] }|d SrNrGrHs r#rKzHYolosImageProcessorPil.post_process_object_detection..rLr%r.rOrQ)rrUrVrWr6rrrXrr rYrZr[r\r]r^r_r`rarbrcs r#rrz4YolosImageProcessorPil.post_process_object_detections& $ ***&~w/AH  #:#l"3"333 j}$$Z44c3B3h++B//)22  #,-- 6 %A%AL%A%A%ABB %A%AL%A%A%ABB+22155 u UE5%$@aHHHKKELYYIIaaaqqqj11E66511 M MGAq!a)m$Ea)m$EA M"C NNeusKK L L L Lr%c tdrtrurws r#rxz9YolosImageProcessorPil.post_process_instance_segmentationryr%c tdr{rurws r#r|z9YolosImageProcessorPil.post_process_semantic_segmentationr}r%c tdrrurws r#rz9YolosImageProcessorPil.post_process_panoptic_segmentationr}r%rr)rrrnpndarrayr rr8rr rZrrrrxr|rrrs@r#rrs 48 44z44/0 4  444444nY]44"'4=G$u+=U4444lnnnwwwwwwwwwwr%r)Nr) typingrnumpyrr[rtorchvision.transforms.v2rtvF.transformers.models.detr.image_processing_detrr2transformers.models.detr.image_processing_pil_detrrimage_transformsr image_utilsr r r utilsr rr get_loggerrloggerrrr$r'r__all__rGr%r#rs 777777MMMMMMTTTTTT888888\\\\\\\\\\;;;;;;;;;;  H % %Z\&&c3h&'*&69Dj&SV& 38_&&&&Rjwjwjwjwjw,jwjwjwZtwtwtwtwtw2twtwtwn !": ;r%