First of all, thank you for providing this dataset and SDK!
I am trying to evaluate various visual odometry methods using this dataset, however, I will need to transform the GPU & IMU to KITTI Pose Ground truth format (from https://www.cvlibs.net/datasets/kitti/eval_odometry.php). Any clues or tools for doing so? Thank you!
Every row of the file contains the first 3 rows of a 4x4 homogeneous pose matrix (SE(3) matrix) flattened into one line, with each value separated by a space. For example, this pose matrix:
First of all, thank you for providing this dataset and SDK! I am trying to evaluate various visual odometry methods using this dataset, however, I will need to transform the GPU & IMU to KITTI Pose Ground truth format (from https://www.cvlibs.net/datasets/kitti/eval_odometry.php). Any clues or tools for doing so? Thank you!
Explanation of the format from https://github.com/MichaelGrupp/evo/wiki/Formats#kitti---kitti-dataset-pose-format
Every row of the file contains the first 3 rows of a 4x4 homogeneous pose matrix (SE(3) matrix) flattened into one line, with each value separated by a space. For example, this pose matrix:
will appear in the file as the row:
a b c d e f g h i j k l