Args: boxes (Boxes, RotatedBoxes or ndarray): either a :class:`Boxes`, or an Nx4 numpy array of XYXY_ABS format f

Normalize, pad and batch the input images.

r""" apply max pooling to get the same effect of nms Args: fmap(Tensor): output tensor of previous step pool_size(int):

r""" apply max pooling to get the same effect of nms Args: fmap(Tensor): output tensor of previous step pool_

Args: cfg (CfgNode): model (nn.Module): evaluators (list[DatasetEvaluator] or None): if None, will ca

get top K point in score map

Load weights from a checkpoint. ckpt trained on this detectron2 framework: checkpoint_state_dict['model'] Offici

L1 regression loss Arguments: regr (batch x max_objects x dim) gt_regr (batch x max_objects x dim) mask (batch x max_objects)

Args: loss_dict (dict): dict of scalar losses data_time (float): time taken by the dataloader iteration

Add config for MOT dataset.

When the config is defined for certain number of workers (according to ``cfg.SOLVER.REFERENCE_WORLD_SIZE``) that's different from the

Build the whole model architecture, defined by ``cfg.MODEL.META_ARCHITECTURE``. Note that it does not load any weights from ``cfg``.

Build the whole model architecture, defined by ``cfg.MODEL.META_ARCHITECTURE``. Note that it does not load any weights from ``cfg``.

Returns: DatasetEvaluator or None It is not implemented by default.

Build a list of default hooks, including timing, evaluation, checkpointing, lr scheduling, precise BN, writing events. Retur

It now calls :func:`detectron2.solver.build_lr_scheduler`. Overwrite it if you'd like a different scheduler.

Build an optimizer from config.

Suppose the number of previous boxes (t1) is dynamic so the boxes in the batches are listed. In contrast, the number of current boxes (t2)

Returns: iterable It now calls :func:`detectron2.data.build_detection_test_loader`. Overwrite it if you'd like a

Build the whole model architecture, defined by ``cfg.MODEL.META_ARCHITECTURE``. Note that it does not load any weights from ``cfg``.

Build the whole model architecture, defined by ``cfg.MODEL.META_ARCHITECTURE``. Note that it does not load any weights from ``cfg``.

Build a list of writers to be used. By default it contains writers that write metrics to the screen, a json file, and a tenso

r""" decode output feature map to detection results Args: fmap(Tensor): output feature map hm(Tensor): output

Args: box_coord (tuple): a tuple containing x0, y0, x1, y1 coordinates, where x0 and y0 are the coordinates of th