Build multiple branch.

Make causal mask used for bi-directional self-attention.

Build fuse layer.

Make each layer.

Build one branch.

Make stem layer for ResNet.

Store the minibach in 'self.batch'

Reference: https://people.sc.fsu.edu/~jburkardt/presentations /truncated_normal.pdf

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

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

Parse iterable values in the string. All elements inside '()' or '[]' are treated as iterable values. Args: val (str): V

Parse the raw outputs (losses) of the network. Args: losses (dict): Raw output of the network, which usually contain

Positiong embeding method. Resize the pos_embed, if the input image size doesn't match the training size. Args:

Substitute base variable placehoders to string, so that parsing would work.

Randomly sample an img_scale according to ``ratio_range`` and ``multiscale_mode``. If ``ratio_range`` is specified, a ratio will be s

Register a handler for some file extensions. Args: handler (:obj:`BaseFileHandler`): Handler to be registered. file_formats (str

Adapted from Mesh Tensorflow: https://github.com/tensorflow/mesh/blob/0cb87fe07da627bf0b7e60475d59f95ed6b5be3d/mesh_tensorflow/transf

Load the best checkpoint for evaluation.

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

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

Saves module state to `destination` dictionary. This method is modified from :meth:`torch.nn.Module._save_to_state_dict`. Args: modu

Saves module state to `destination` dictionary. This method is modified from :meth:`torch.nn.Module._save_to_state_dict`. Args: modu

Substitute variable strings to their actual values.

R4R is not the shortest path. The goal location can be visited nodes.

Convert image based absolute point coordinates to image based relative coordinates for sampling. Args: abs_img_points (Tensor): Image

Add all parameters of module to the params list. The parameters of the given module will be added to the list of param groups, with s

Called after every training epoch to evaluate the results.

Called after every training iter to evaluate the results.

Placeholder for augmentation test.

Scaling bboxes w.r.t the box center. Args: bboxes (ndarray): Shape(..., 4). scale (float): Scaling factor. clip_shape (tu

initialize conv/fc bias value according to a given probability value.

Given an input and a flow-field grid, computes the output using input values and pixel locations from grid. Supported only bilinear interpolation

Brightness distortion.

Create by split datasets inheriting torch.utils.data.Datasets. # build() can be dataset-specific. Overwrite to customize.

Builder for attention.

Build a dictionary of datasets, keyed by split 'train', 'valid', 'test'. Download dataset and annotations automatically if not exist.

Builder for feed-forward network (FFN).

Build padding layer. Args: cfg (None or dict): The padding layer config, which should contain: - type (str): Layer type.

Build pixel sampler for segmentation map.

Build plugin layer. Args: cfg (None or dict): cfg should contain: type (str): identify plugin layer type. layer a

NOTE: shape arg should include spatial dim only