Args: cfg (CfgNode):

Evaluate detection proposal recall metrics. This function is a much faster alternative to the official COCO API recall evaluation code. Howev

Get COCO API "tasks" (i.e. iou_type) from COCO-format predictions.

Accumulate per image evaluation results and store the result in self.eval :param p: input params for evaluation :return: None

Add config for MASK_FORMER.

xyz: B X V X H X W X 3

Inputs: multi_scale_features: list [B*V, 256, 15, 20], [B*V, 256, 30, 40], [B*V, 256, 60, 80] depths: tensor [B, 5, 4

input: grad_out: (m, c, nsample) output: (n, c), None

================================== ballquery_batch_p ================================== */ input xyz: (n, 3) float input batch_idxs: (n) int input bat

input: semantic_label, int, N input: ball_query_idxs, int, (nActive) input: start_len, int, (N, 2) output: cluster_idxs, int (sumNPoint, 2), dim 0 for

Build a batched dataloader. The main differences from `torch.utils.data.DataLoader` are: 1. support aspect ratio grouping options 2. use

Build a dataloader for object detection with some default features. This interface is experimental. Args: dataset (list or torch

Create evaluator(s) for a given dataset. This uses the special metadata "evaluator_type" associated with each builtin dataset.

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

Build a pixel decoder from `cfg.MODEL.MASK_FORMER.PIXEL_DECODER_NAME`.

Create a ResNet instance from config. Returns: ResNet: a :class:`ResNet` instance.

Create a list of default :class:`Augmentation` from config. Now it includes resizing and flipping. Returns: list[Augmentation]

Create a list of default :class:`Augmentation` from config. Now it includes resizing and flipping. Returns: list[Augmentation]

Create a list of default :class:`Augmentation` from config. Now it includes resizing and flipping. Returns: list[Augmentation]

Build a instance embedding branch from `cfg.MODEL.INS_EMBED_HEAD.NAME`.

Calculates the principal components given the embedding features. Args: embeddings: A 2-D float tensor of shape `[num_pixels, embedding_

We estimate uncerainty as L1 distance between 0.0 and the logit prediction in 'logits' for the foreground class in `classes`. Args:

:param batch_idxs: (N), int, cuda :labels: 0-19

Collate function for training. is called ONLY in train loader Args: batch (list[dict]): a list of samples, each sample is a dict of i

Here the logits are raw outputs not the softmax output! logits: (batch_size, q, seq_len) num_class: N positive_map_od: (batch_size, s

Create a DistributedDataParallel model if there are >1 processes. Args: model: a torch.nn.Module fp16_compression: add fp16

construct a map such that positive_map[i,j] = True iff box i is associated to token j

construct a map such that positive_map[i] = j, where j is the object detection label of the token i

Download COCO images from mscoco.org server. :param tarDir (str): COCO results directory name imgIds (list): images to

Download files for a specifed scan_id.

Args: img_ids: a list of image IDs to evaluate on. Default to None for the whole dataset

Args: img_ids: a list of image IDs to evaluate on. Default to None for the whole dataset