Print information about the annotation file. :return:

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

Run all_gather on arbitrary picklable data (not necessarily tensors) Args: data: any picklable object Returns: list[data]

Load and prepare dataset dicts for instance detection/segmentation and semantic segmentation. Args: dataset_names (str or list[str])

Get ann ids that satisfy given filter conditions. default skips that filter :param imgIds (int array) : get anns for given imgs

Inputs: source_feats: tensor [B*V, C, H1, W1] or B, N, C source_xyz: tensor [B, V, H1, W1, 3] or B, N, 3 source_p2v: tens

Load anns with the specified ids. :param ids (int array) : integer ids specifying anns :return: anns (object array) : l

Create configs and perform basic setups.

Args: stem (nn.Module): a stem module stages (list[list[CNNBlockBase]]): several (typically 4) stages,

Return an activation function given a string

inpaints depth using opencv Input: torch tensor with depthvalues: H, W Output: torch tensor with depthvalues: H, W

Create a list of blocks of the same type that forms one ResNet stage. Args: block_class (type): a subclass of CNNBlockBa

Derive the desired score numbers from summarized COCOeval. Args: coco_eval (None or COCOEval): None represents no predic

Evaluate the coco results using COCOEval API.

A context which ignores CUDA OOM exception from pytorch.

Prepare ._gts and ._dts for evaluation based on params :return: None

Takes in all scannet images, and add its multiview images as contexts.

Convert annotation which can be polygons, uncompressed RLE to RLE. :return: binary mask (numpy 2D array)

Compute the DICE loss, similar to generalized IOU for masks Args: inputs: A float tensor of arbitrary shape. The pred

Similar to `build_detection_train_loader`, but uses a batch size of 1. This interface is experimental. Args: dataset (list or to

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

Convert object detection data into grounding data format, on the fly. ind_to_class: {0: "__background__", 1 : "person" ...}, contiguous id

Project high-dimension embeddings to RGB colors. Args: embeddings: A 3-D float tensor with shape `[batch_size, embedding_dim, num_

Freeze the first several stages of the ResNet. Commonly used in fine-tuning. Layers that produce the same feature map spatia

filtering parameters. default skips that filter. :param catNms (str array) : get cats for given cat names :param supNms (str

Get img ids that satisfy given filter conditions. :param imgIds (int array) : get imgs for given ids :param catIds (int array

takes in a video dataset and retuns dict of scene to images mapping returns : dict{scene_id: str, image_dataset_ids: []}

Dump an "Instances" object to a COCO-format json that's used for evaluation. Args: instances (Instances): img_id (int): the

Args: inputs['all_instances']: all k-frames instances in detectron2 format Returns: [{ "masks": list of g

Conducts principal component analysis on the embedding features. This function is used to reduce the dimensionality of the embedding. Args:

Use kill(0) to determine if pid is alive :param pid: process id :rtype: bool

Compute and display summary metrics for evaluation results. Note this functin can *only* be applied on the default parameter setting

Compute and display summary metrics for evaluation results given a custom value for max_dets_per_image

Inputs: xyz: tensor [B, N, 3] voxel_size: float Outputs: point_to_voxel_all: tensor [B, N], is the mapping from orig

Args: x: (B, H, W, C) window_size (int): window size Returns: windows: (num_windows*B, window_size, window_size, C)