I’m currently working on a visual odometry project, and I don’t have a ZED Camera now.
Is it possible that I can input lots of pairs of stereo images so I can first decide weather to buy a ZED camera based on the testing result?
Also, from the Stereo Visual Odometry document page
It says that this project can be used to evaluate the odometry of Kitti dataset.
How to reproduce this Kitti evaluation experiment using this NVidia Isaac Stereo Visual Odometry sample?
Thank you very much!
You can write a custom codelet to read a bunch of pairs of stereo images and pass them accordingly to the StereoVisualOdometry codelet. You have to provide a left image, right image, and camera extrinsic to the StereoVisualOdometry codelet. Details about that can be found at the following link
I’m just curious is there a document that teach us how to reproduce the kitti dataset experiment?