Run forward function and calculate loss for point head in training.

Export mask refining process with point head to onnx. Args: x (tuple[Tensor]): Feature maps of all scale level. rois

Export the process of sampling fine grained feats to onnx. Args: x (tuple[Tensor]): Feature maps of all scale level.

Pad images according to ``self.size``.

Pad masks according to ``results['pad_shape']``.

Pad semantic segmentation map according to ``results['pad_shape']``.

Parse bbox and mask annotation. Args: ann_info (list[dict]): Annotation info of an image. with_mask (bool): Whether t

Parse bbox and mask annotation. Args: ann_info (list[dict]): Annotation info of an image. with_mask (bool): Whether t

Private function to convert masks represented with polygon to bitmaps. Args: mask_ann (list | dict): Polygon mask annotat

Compute the area of a component of a polygon. Using the shoelace formula: https://stackoverflow.com/questions/24467972/calculate-area

Convert proposal results to COCO json style.

Convert proposal results to COCO json style.

Get another random index from the same group as the given index.

Ramdom jitter positive proposals for training.

Resize bounding boxes with ``results['scale_factor']``.

Resize images with ``results['scale']``.

Resize masks with ``results['scale']``

Resize semantic segmentation map with ``results['scale']``.

Rotate the bboxes.

Rotate the image. Args: results (dict): Result dict from loading pipeline. angle (float): Rotation angle in degrees,

Rotate the masks.

Rotate the segmentation map.

Sample negative samples.

Sample negative samples. Score-HLR sampler is done in the following steps: 1. Take the maximum positive score prediction of each nega

Sample positive samples.

Randomly sample some positive samples.

Convert instance segmentation results to COCO json style.

Convert instance segmentation results to COCO json style.

Set flag according to image aspect ratio. Images with aspect ratio greater than 1 will be set as group 1, otherwise group 0.

Shear the bboxes.

Shear the image. Args: results (dict): Result dict from loading pipeline. magnitude (int | float): The magnitude used

Shear the masks.

Shear the segmentation maps.

Get features from pos rois.

Around padding the original image without cropping. The padding mode and value are from ``test_pad_mode``. Args: results

Random crop and around padding the original image. Args: results (dict): Image infomations in the augment pipeline. Retu

Shift bboxes horizontally or vertically, according to offset.

Translate the image. Args: results (dict): Result dict from loading pipeline. offset (int | float): The offset for tr

Translate masks horizontally or vertically.

Translate segmentation maps horizontally or vertically.

Adjusts the compatibility of widths and groups. Args: widths (list[int]): Width of each stage. bottleneck_ratio (floa

Associative Embedding Loss in one image. Associative Embedding Loss including two parts: pull loss and push loss. Pull loss makes embedding v

Assign only one gt index to each prior box. Gts with large gt_priority are more likely to be assigned. Args: is_bbox_in_

Asynchronized test for box head without augmentation.

Asynchronized test for mask head without augmentation.

Extract direction-specific features fx and fy with attention methanism.

Test function with test time augmentation.

Test det bboxes with test time augmentation.

Test for mask head with test time augmentation.

Calculate average precision (for single or multiple scales). Args: recalls (ndarray): shape (num_scales, num_dets) or (num_dets, )

Calculate balanced L1 loss. Please see the `Libra R-CNN <https://arxiv.org/pdf/1904.02701.pdf>`_ Args: pred (torch.Tensor): The pred

Generate buckets estimation and fine regression targets. Args: proposals (Tensor): Shape (n, 4) gt (Tensor): Shape (n, 4)

Compute deltas of proposals w.r.t. gt. We usually compute the deltas of x, y, w, h of proposals w.r.t ground truth bboxes to get regression t

Flip bboxes horizontally. Args: bboxes (numpy.ndarray): Bounding boxes, shape (..., 4*k) img_shape (tuple[int]): Imag

Export bbox branch to onnx which supports batch inference. Args: x (tuple[Tensor]): Feature maps of all scale level.

Encode ground truth boxes to tblr coordinate. It first convert the gt coordinate to tblr format, (top, bottom, left, right), relative to prio

Encode ground truth boxes to tblr coordinate. It first convert the gt coordinate to tblr format, (top, bottom, left, right), relative to pri

Compute the area of an array of bboxes. Args: bboxes (Tensor): The coordinates ox bboxes. Shape: (m, 4) Returns: Tensor: Are

BIoULoss. This is an implementation of paper `Improving Object Localization with Fitness NMS and Bounded IoU Loss. <https://arxiv.org/abs

Apply bucketing estimation (cls preds) and fine regression (offset preds) to generate det bboxes. Args: proposals (Tensor): Boxes to

Build a module. Args: cfg (dict, list[dict]): The config of modules, is is either a dict or a list of configs. registr

Build distill loss.

Build RoI operator to extract feature from each level feature map. Args: layer_cfg (dict): Dictionary to construct and config RoI

Build shared head.

Compute point specific representation regions. See Grid R-CNN Plus (https://arxiv.org/abs/1906.05688) for details.

Calculate positive recall with score threshold. Args: cls_scores (list[Tensor]): Classification scores at all fpn levels.

Compute centerness targets. Args: pos_bbox_targets (Tensor): BBox targets of positive bboxes in shape (num_pos, 4

Get bboxes according to center points. Only used in :class:`MaxIoUAssigner`.

r"""`Implementation of paper `Enhancing Geometric Factors into Model Learning and Inference for Object Detection and Instance Segmentation <ht

:meth: match the detection results with the groundtruth by Caltech matching strategy :param thres: iou threshold :type thres:

:meth: match the detection results with the groundtruth by VOC matching strategy :param thres: iou threshold :type thres: flo