cvtkit.io

Pose File Formats

Please see the resources page on pose file formats to learn more about the expected/supported formats commonly used in this library.

A suite of common input/output functions.

This module includes several functions for reading and writing different types of data useful for computer vision applications.

load_pretrained_model(model, ckpt)

Loads model weights from disk.

read_cams_sfm(camera_path, extension='cam.txt')

Reads an entire directory of camera files in SFM format.

Parameters:

Name	Type	Description	Default
camera_path	str	Path to the directory of camera files.	required
extension	str	File extension being used for the camera files.	'cam.txt'

Returns:

Type	Description
ndarray	Array of camera extrinsics, intrinsics, and view metadata (Nx2x4x4).

read_cams_trajectory(log_file)

Reads camera file in Trajectory File format.

Parameters:

Name	Type	Description	Default
log_file	str	Input *.log file to be read.	required

Returns:

Type	Description
NDArray[float32]	Array of camera extrinsics, intrinsics, and view metadata (Nx2x4x4).

read_cluster_list(filename)

Reads a cluster list file encoding supporting camera viewpoints.

Parameters:

Name	Type	Description	Default
filename	str	Input file encoding per-camera viewpoints.	required

Returns:

Type	Description
List[Tuple[int, List[int]]]	An array of tuples encoding (ref_view, [src_1,src_2,..])

read_extrinsics_tum(tum_file, key_frames=None)

Reads extrinsic camera trajectories in TUM format [timestamp tx ty tz qx qy qz qw].

Parameters:

Name	Type	Description	Default
tum_file	str	Input extrinsics file.	required
key_frames	List[int] | None	Indices corresponding to the desired keyframes.	None

Returns:

Type	Description
ndarray	Array of camera extrinsics (Nx4x4).

read_matrix(mat_file)

Reads a single matrix of float values from a file.

Parameters:

Name	Type	Description	Default
mat_file	str	Input file for the matrix to be read.	required

Returns:

Type	Description
ndarray	The matrix stored in the given file.

read_mesh(mesh_file)

Reads a mesh from a file.

Parameters:

Name	Type	Description	Default
mesh_file	str	Input mesh file.	required

Returns:

Type	Description
TriangleMesh	The mesh stored in the given file.

read_pfm(pfm_file)

Reads a file in *.pfm format.

Parameters:

Name	Type	Description	Default
pfm_file	str	Input *.pfm file to be read.	required

Returns:

Type	Description
NDArray[float32]	Data map that was stored in the *.pfm file.

read_point_cloud(point_cloud_file)

Reads a point cloud from a file.

Parameters:

Name	Type	Description	Default
point_cloud_file	str	Input point cloud file.	required

Returns:

Type	Description
PointCloud	The point cloud stored in the given file.

read_point_cloud_np(point_cloud_file)

Reads a point cloud from a file.

Parameters:

Name	Type	Description	Default
point_cloud_file	str	Input point cloud file.	required

Returns:

Type	Description
NDArray[float32]	The point cloud as an NDArray stored in the given file.

read_single_cam_sfm(cam_file, depth_planes=256)

Reads a single camera file in SFM format.

Parameters:

Name	Type	Description	Default
cam_file	str	Input camera file to be read.	required
depth_planes	int	Number of depth planes to store in the view metadata.	256

Returns:

Type	Description
ndarray	Camera extrinsics, intrinsics, and view metadata (2x4x4).

read_stereo_intrinsics_yaml(intrinsics_file)

Reads intrinsics information for a stereo camera pair from a *.yaml file.

Parameters:

Name	Type	Description	Default
intrinsics_file	str	Input *.yaml file storing the intrinsics information.	required

Returns:

Name	Type	Description
K_left	NDArray[float32]	Intrinsics matrix (3x3) of left camera.
D_left	NDArray[float32]	Distortion coefficients vector (1x4) of left camera.
K_right	NDArray[float32]	Intrinsics matrix (3x3) of right camera.
D_right	NDArray[float32]	Distortion coefficients vector (1x4) of right camera.
R	NDArray[float32]	Relative rotation matrix (3x3) from left -> right cameras.
T	NDArray[float32]	Relative translation vector (1x3) from left -> right cameras.

sfm_to_trajectory(cams, log_file)

Convert a set of cameras from SFM format to Trajectory File format.

Parameters:

Name	Type	Description	Default
cams	NDArray[Any] | Tensor	Array of camera extrinsics (Nx4x4) to be converted.	required
log_file	str	Output path to the *.log file that is to be created.	required

trajectory_to_sfm(log_file, camera_path, intrinsics)

Convert a set of cameras from Trajectory File format to SFM format.

Parameters:

Name	Type	Description	Default
log_file	str	Input *.log file that stores the trajectory information.	required
camera_path	str	Output path where the SFM camera files will be written.	required
intrinsics	ndarray	Array of intrinsics matrices (Nx3x3) for each camera.	required

write_cam_sfm(cam_file, intrinsics, extrinsics)

Writes intrinsic and extrinsic camera parameters to a file in sfm format.

Parameters:

Name	Type	Description	Default
cam_file	str	The file to be writen to.	required
intrinsics	ndarray	Camera intrinsic data to be written.	required
extrinsics	ndarray	Camera extrinsic data to be written.	required

write_matrix(M, mat_file)

Writes a single matrix to a file.

Parameters:

Name	Type	Description	Default
M	ndarray	Matrix to be stored.	required
mat_file	str	Output file where the given matrix is to be writen.	required

write_mesh(mesh_file, mesh)

Writes a mesh to a file.

Parameters:

Name	Type	Description	Default
mesh_file	str	Output mesh file.	required
mesh	TriangleMesh	Mesh to be stored.	required

write_pfm(filename, data_map, scale=1.0)

Writes a data map to a file in *.pfm format.

Parameters:

Name	Type	Description	Default
pfm_file		Output *.pfm file to store the data map.	required
data_map	ndarray	Data map to be stored.	required
scale	float	Value used to scale the data map.	1.0