Build loss.

Creates a data loader with the specified parameters. Args: dataset: A dataset (third party, LaViDa or WebDataset). batch_siz

Dump model and checkpointables to a file. Args: name (str): name of the file. kwargs (dict): extra arbitrary

Setup logging. Args: output: A file name or a directory to save log files. If None, log files will not be saved. If outp

Wrapper for grid_sample, uses pixel coordinates

3x3 convolution with padding

Description: Generates cost-volume using left image features, disaprity samples a

Warping by disparity Args: img: [B, 3, H, W] disp: [B, 1, H, W], positive padding_mode: 'zeros' or 'border' Returns:

Inference with slide/whole style. Args: img (Tensor): The input image of shape (N, 3, H, W). img_meta (dict): Image i

Returns: True if distributed training is enabled

Args: rng (None | int | numpy.random.RandomState): seed or state. kwargs (keyword arguments): - num_p

fmap: [B, C, H, W] disp_sample: tensor of dim [B*H*W, num_disp], disparity samples radius(int): 2*radius+1 samples will be sa

Create configs and perform basic setups.

Return an activation function given a string

Add prefix for dict. Args: inputs (dict): The input dict with str keys. prefix (str): The prefix to add. Returns: d

Build Match Cost.

Encode images with backbone and decode into a semantic segmentation map of the same size as input.

Encode images with backbone and decode into a depth estimation map of the same size as input.

Encode images with backbone and decode into a depth estimation map of the same size as input.

Extract features from images.

Forward function for training. Args: img (Tensor): Input images. img_metas (list[dict]): List of image info dict wher

coord: [...]D returns: [...]dim, where dim=(2*N_freqs+1)*D

Generating attention mask to prevent message passing along self edges. Args: window_size (tuple[int]): The height, width