Utils Reference

MultiViewTracks.utils.compute_reprojection_errors(tracks, tracks_reprojected, identities=[])

Computes point-wise reprojection errors (distances) between tracks and their reprojections.

Parameters:
  • tracks (dict) – A dictionary containing the original tracks
  • tracks_reprojected (dict) – A dictionary containing the reprojected tracks
  • identities (list, optional) – A list of trajectory identities for which the reprojection errors should be computed
Returns:

  • list – A list of arrays containing the reprojection errors for each identity
  • array – An array containing the respective identities
MultiViewTracks.utils.load(file_name)

Loads a python object from a .pkl file

Parameters:file_name (str) – File path of saved object
Returns:Loaded python object
Return type:object
MultiViewTracks.utils.plot_tracks_2d(tracks, ax=None, figsize=(30, 30), show=True, style='scatter', size=1.0)

Plots the x and y components of tracks. Can be used to visualize reprojection errors.

Parameters:
  • tracks (dict) – A dictionary containing the tracks
  • ax (matplotlib.pyplot.Axes, optional) – Axes for plotting
  • figsize ((int, int), optional) – Size of the matplotlib output if ax is not specified. Defaults to (30, 30)
  • show (bool, optional) – Show plot calling plt.show. Defaults to True.
  • style (str, optional) – Plot style, one of “line”, “scatter” or “errors”. Defaults to “scatter”.
  • size (float, optional) – Line width or marker size. Defaults to 1.0.
Returns:
Return type:

matplotlib.pyplot.Axes
MultiViewTracks.utils.pointcloud_to_ply(point_cloud)

Returns pre-formatted ply points from input points, use Scene.get_pointcloud

MultiViewTracks.utils.read_next_bytes(fid, num_bytes, format_char_sequence, endian_character='<')

Read next bytes of a COLMAP .bin file.

See https://github.com/colmap/colmap/blob/dev/scripts/python/read_model.py for reference.

MultiViewTracks.utils.save(dump, file_name)

Save to a .pkl file

Parameters:
  • dump (object) – Python object to save
  • file_name (str) – File path of saved object
Returns:

Successful save?
Return type:

bool
MultiViewTracks.utils.tracks_to_ply(tracks, uniform_color=None)

Prepare tracks for ply file save.

Parameters:
  • tracks (dict) – A tracks dictionary, not a pooled dictionary, must contain z component
  • uniform_color ((int, int, int), optional) – A shared color used for all tracks, otherwise random RGB generation
Returns:

A list of lists of per individual pre-formatted ply points (x y z r g b a)
Return type:

list
MultiViewTracks.utils.triangulate_point(pts_2d, views)

Triangulate points from multiple views using either OpenCV.triangulatePoints or DLT.

See https://github.com/opencv/opencv_contrib/blob/master/modules/sfm/src/triangulation.cpp for reference.

MultiViewTracks.utils.write_ply(pts_ply, file_name)

Write points to a .ply file for visualization

Parameters:
  • pts_ply (list) – The prepared points in ply format, for example using tracks_to_ply
  • file_name (str) – The file name of the saved file
Returns:

Successful save?
Return type:

bool