NgU xdZddlZddlmZdejfdZ dd ed ed ed efd Z dd ed ed ed efdZ dS)z PyTorch Soft-NMS This code was adapted from a PR for detectron2 submitted by https://github.com/alekseynp https://github.com/facebookresearch/detectron2/pull/1183/files Detectron2 is licensed Apache 2.0, Copyright Facebook Inc. N)Listreturnc ||dddf|dddfz |dddf|dddfz z}|dddf|dddfz |dddf|dddfz z}tj|dddddf|ddddftj|dddddf|ddddfz }|d|d}tj|dk||dddf|z|z z tjd|j|j}|S) aH Given two lists of boxes of size N and M, compute the IoU (intersection over union) between __all__ N x M pairs of boxes. The box order must be (xmin, ymin, xmax, ymax). Args: boxes1,boxes2 (Boxes): two `Boxes`. Contains N & M boxes, respectively. Returns: Tensor: IoU, sized [N,M]. Nr)min)dimdtypedevice) torchr maxclamp_prodwherezerosr r )boxes1boxes2area1area2 width_heightinterious K/var/www/html/ai-engine/env/lib/python3.11/site-packages/effdet/soft_nms.py pairwise_iour sAAAqD\F111a4L (VAAAqD\F111a4L-H IE AAAqD\F111a4L (VAAAqD\F111a4L-H IE9VAAAtQRRK0&ABB-@@59qqq${VAAArrE]DDLA   !  $ $E +   qqq$w%'%/0 AU[>>>  C JT?{Gzt?method_gaussiansigma iou_thresholdscore_thresholdc|j}|}|}|d} tj| } tj| tj|} tj| tj|} d} |dkr tj |}| || | <||| | <| dz } ||}t| d|d}|r,tj tj |d |z }n(tj|}||k}d||z ||<||z}||k}tjd|||<||}||}| |} |dk | d| | d| fS)a Soft non-max suppression algorithm. Implementation of [Soft-NMS -- Improving Object Detection With One Line of Codec] (https://arxiv.org/abs/1704.04503) Args: boxes_remain (Tensor[N, ?]): boxes where NMS will be performed if Boxes, in (x1, y1, x2, y2) format if RotatedBoxes, in (x_ctr, y_ctr, width, height, angle_degrees) format scores_remain (Tensor[N]): scores for each one of the boxes method_gaussian (bool): use gaussian method if True, otherwise linear sigma (float): parameter for Gaussian penalty function iou_threshold (float): iou threshold for applying linear decay. Nt from the paper re-used as threshold for standard "hard" nms score_threshold (float): boxes with scores below this threshold are pruned at each iteration. Dramatically reduces computation time. Authors use values in [10e-4, 10e-2] Returns: tuple(Tensor, Tensor): [0]: int64 tensor with the indices of the elements that have been kept by Soft NMS, sorted in decreasing order of scores [1]: float tensor with the re-scored scores of the elements that were kept rr rrF)r N)r clonesizerarangerint64float32numelargmaxr unsqueezeexppow ones_liketensor)boxesscoresr r!r"r#r boxes_remain scores_remainnum_elemidxsidxs_out scores_outcounttop_idxtop_boxiousdecay decay_maskkeeps rsoft_nmsr@*sJ\F;;==LLLNNM!!##A&H < ! !D{85;vFFFHXU]6JJJJE     ! # #,}--w-)'2 5  w'G--a00,??B  5Iuyq111E9::EEOD))E -J !D$4 4E*  . U6:::W #D) %d+ Dz-     ! # #0 FUF Z/ //rMbP?cp|dkrLtjdtj|jtjdtj|jfS|}|||dzz}||dddfz} t| |||||S)a Performs soft non-maximum suppression in a batched fashion. Each index value correspond to a category, and NMS will not be applied between elements of different categories. Args: boxes (Tensor[N, 4]): boxes where NMS will be performed. They are expected to be in (x1, y1, x2, y2) format scores (Tensor[N]): scores for each one of the boxes idxs (Tensor[N]): indices of the categories for each one of the boxes. method (str): one of ['gaussian', 'linear', 'hard'] see paper for details. users encouraged not to use "hard", as this is the same nms available elsewhere in detectron2 sigma (float): parameter for Gaussian penalty function iou_threshold (float): iou threshold for applying linear decay. Nt from the paper re-used as threshold for standard "hard" nms score_threshold (float): boxes with scores below this threshold are pruned at each iteration. Dramatically reduces computation time. Authors use values in [10e-4, 10e-2] Returns: tuple(Tensor, Tensor): [0]: int64 tensor with the indices of the elements that have been kept by Soft NMS, sorted in decreasing order of scores [1]: float tensor with the re-scored scores of the elements that were kept r)rr rN)r r!r"r#) r*remptyr(r r)rtor@) r1r2r6r r!r"r#max_coordinateoffsets boxes_for_nmss rbatched_soft_nmsrHssN {{}} KEK E E E KEM&- H H H  YY[[Nggenn 23GGAAAtG,,M ve#_   r)Trrr)TrrrA) __doc__rtypingrTensorrboolfloatr@rHrrrOs ELB!" F0F0F0  F0  F0  F0F0F0F0V!" 666 6 6  666666r