Returns the number of files in the index file.

Return an estimation of the number of batches per epoch.

Get the xyz_w contained in the dataset in meters.

Register a figure to the tensorboard. See tensorboard's documentation https://pytorch.org/docs/stable/tensorboard.html for usage.

Register a histogram to the tensorboard. See tensorboard's documentation https://pytorch.org/docs/stable/tensorboard.html for usage.

Given an image, pad the image at the top and add a title text. Note that the function converts image to uint8 Args: image:

Baseline point cloud-ray intersection. Algorithm: 1. poisson surface reconstruction from point cloud to create a mesh 2. ray tracing

concatenate at `dim`.

Concatenate a list of PointersectRecord at the given dimension. It is useful to split m_shape.

Chunk the PointersectRecord at the given dimension. As pytorch, the resulted tensors are views to the original ones.

Clean up glb such that open3d can load the texture maps. Note that the function creates a gtlf file (with the same name) from the glb fil

Given ray2gidx and gidx2pidx, construct ray2pidxs. Args: ray2gidx: (b, m, n_gidxs), gidx can be outside the grid n_ray2gidx: (b, m), number of neighb

Compare a list of rgbd_images by creating gif containing each of the rgbd_images Note we assume the camera used to capture the rgbd images ar

Given a config_dict: key -> val, returns a list of ["--key", "val"] so that argparse can parse it. Args: config_dict:

Compose the command to execute on bolt. Args: script_filename: the .py filename. Note that it is assumed to be executed

Compute the xyz in the world coordinate using z_map and camera pose. Important note: The function uses a image coordinate system: x

Compute the zdir and dps in the world coordinate using z_map and camera pose. Args: z_map: (*, h, w) the z coordinate of

Compute the accuracy. Args: arr: (*b_shape, *d_shape), binary ref: (*b_shape, *d_shape), binary ndim_b:

Compute the area spanned by the unit vectors in arr and ref. Args: arr: (*b_shape, *d_shape, 3) ref: (*b_shape, *d_shape, 3)

Compute the angle spanned by the unit vectors in arr and ref. Args: arr: (*b_shape, *d_shape, 3) ref: (*b_shape, *d_shape, 3

Compute average l1 distance between arr and ref. Args: arr: (*b_shape, *d_shape) ref: (*b_shape, *d_shape) ndim_b:

Compute the center coordinate of the first feature of each layer. Note that the function assumes symmetric padding: for example, inp

Compute the sequence length of the output of each layers, with no padding (valid). Args: seq_len_in (int):

Compute the LPIPS metric. Args: arr: (*b_shape, *d_shape) Assumes the RGB image is in [0,1] ref: (*b_shape, *d_shape) Assum

Compare a list of rgbd_images, create gif, compute difference from ref_rgbd_image if given. Note we assume the camera used to capture the rgb

Compute the negative log-likelihood given the samples. Args: samples: (*m_shape, 3) one for each independ

Compute the output sequence length. Args: pad_type: ``None``: use the original padding config ``'

Compute the output sequence length. Args: pad_type: ``None``: use the original padding config ``'

Compute the distance between each point to each ray. Args: points: (*, n, 3) ray_origins: (*, m, 3)

Compute peak signal to noise ratio Args: arr: (*b_shape, *d_shape) ref: (*b_shape, *d_shape) ndim_b: numb

Compute the root mean squared error between arr and ref. The average is taken over the d_shape. Args: arr: (*b_shape, *d_shape)

If we want to maintain the seq_len during downsampling, every input seq_len to a downsampling layer need to be an even number.

Compute the ssim. Args: arr: (*b_shape, *d_shape) ref: (*b_shape, *d_shape) ndim_b: number of dimension o

Given labels :param labels: (seq_len, batch_size,), (batch_size,), or (*) int, [0, total_classes-1] :param total_classes: total number of

construct time-shifted version of the ground-truth to be used as teacher-forcing input :param gt_vectors: (seq_len, batch, dim_feature) :

Create a gif from the images Args: images: (n, h, w, 3) or list of (h, w, 3), float, range = 0-1 filename:

Create an o3d mesh representing a plane. The mesh contains two triangles. Args: top_left: the top left coordinate of

Store the xys points into an octree. Args: points: (*, 3) max_depth: max depth of the octree. As the

Create a video from the images Args: images: (n, h, w, 3) or list of (h, w, 3), float, range = 0-1 filename:

A decorator indicating methods to be customized.

Derive camera intrinsic matrix. Args: width_px: width (pixel) height_px: height (pixel) fov: field-of-view (degree)

Mimic the behavior of :py:`o3d.visualization.draw_geometries` but free objects properly. Args: geometry_list: List of ge

Set up at the beginning of an epoch, before dataloder iterator is constructed. It can be used to setup the batch sampler, etc.

Compute the correspoding points in image coordinates of the pixels in a source image. Args: uv_c: (*b_shape, *n_shape, 2

Find all the points within ray_radius of a ray. Args: points: (b, n, 3) the xyz_w of points in world coord ray_

Find the k nearest points within `radius` of a ray, ie, the vertical distance from the point to ray <= radius. If not enough points found, return dumm

Modified version of find_neighbor_points_of_rays_brute_force return k neighbors within ray_radius report the valid numbers of neighbors i

Find the k nearest points within `radius` of a ray, ie, the vertical distance from the point to ray <= radius. If not enough points found, return dumm

Find all the points within ray_radius of a ray. Args: points: (b, n, 3) the xyz_w of points in world coord ray_

Find the points within `radius` of a ray, ie, the vertical distance from the point to ray <= radius. Returns: list of list of list:

Find the points within `radius` of a ray, ie, the vertical distance from the point to ray <= radius. Returns: list of list of list: b -> m -> n_idx

Find the points within `radius` of a ray, ie, the vertical distance from the point to ray <= radius. Returns: ray2pidx: (b, M), all the points. The p

Find the k nearest points by each ray and rectify the coordinate system as if the rays are in the direction of (0,0,1) and originated from (0

Fit a hyperplane for the n points such that the plane minimizes the point-to-plane distances. .. math:: \min_{n,c} \sum_i w_i (n^T