transform_poses.py
is a script which transforms the reconstructed model produced
by AliceVision Meshroom into the world frame, for use in the path planning for welding.
For installing basic requirements, see the main README.md. :
To run the standalone script, from transform_poses.py
import
transform_model_to_world_coordinates()
, which takes the following arguments:
- path_to_poses_dir: the path to the directory which contains the raw images and
.npy
camera pose files. - path_to_cameras_sfm: the path to the directory which contains the
cameras.sfm
file, typically located in thecache/*/StructureFromMotion
directory. - path_to_computed_mesh: the path to the directory which contains the
.obj
,.png
and.mtl
files, typically located in thecache/*/Texturing
directory. - path_to_transformed_mesh_dir: the path to the directory where the script should output the transformed mesh files (WARNING: directory should exist).
- show_plot: (Optional) If true, the real camera positions and computed positions, as well as the original and transformed meshes are visualized in a popup.
- exclude_poses: (Optional) A list of strings containing the filenames (without extensions) of the images corresponding to the camera positions that you would like to exclude from the computation.
The script uses the known camera positions provided by the robot, in the world frame, to perform a series of linear transformation (traslation, rotation, scaling) in order to bring the model from the arbitrary frame into the world frame.
The following trasnformations are used:
under construction
- Orfeas Kypris - Initial work - orphefs
See also the list of contributors who participated in this project.
This project is licensed under the MIT License - see the LICENSE.md file for details
- Props to the AliceVision Meshroom development community
- Props to my cats for being cool