Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Evaluation of Pose #10

Open
CSTXBH opened this issue Jan 14, 2022 · 3 comments
Open

Evaluation of Pose #10

CSTXBH opened this issue Jan 14, 2022 · 3 comments

Comments

@CSTXBH
Copy link

CSTXBH commented Jan 14, 2022

Hello,

I want to reproduce the results of pose evaluation with your method. During this process, I was confused by some problems as blow:

  1. Following issue About KITTI Pose #8 , I predicted rel_pose with your model first, transformed them into abs_pose using VO evaluation code provided in your answer, and evaluated them with KITTI odometry evaluation toolbox mentioned in README.md.. But the result is also not so good, closing to results in the issue. I found in the code, that the pose is calculated by RANSAC with default choice, which means just using flow matches to get the pose(is that right?). Is it because the default choice that caused the result not good, or the default choice is enough to get the best result? I was wondering how to set the config to get the best result as in Table 3 of your paper.

  2. As in Table 5, there are many choice to calculate pose. But I don't know how to use them in your code . There are so many flags in config.py, and some of them are dummy. Which flags should I set 'True' to use,for example, the method with best performence in Table 5 as '5-point' + ' Flow matches '+ 'SIFT Loc'?

@jytime can you help me with them? Thank you very much!

Best

@jytime
Copy link
Owner

jytime commented Jan 19, 2022

Hi,

The default choice is the right one, since I have configured the flags in the kitti.yml. It filters the flow matches via SIFT Loc, and computes the pose using the filtered matches. The default one should have been '5-point' + ' Flow matches '+ 'SIFT Loc'.

I notice that some people report their RANSAC CUDA extension does not work well, which might be the reason for the wrong performance. Would you mind sharing your results (relative poses predicted by the network) of sequence 09? I may have a look at the results.

Best,
Jianyuan

@CSTXBH
Copy link
Author

CSTXBH commented Feb 4, 2022

Hi! Thanks! Happy the Spring Festival !

Sorry for late reply. Following steps in issue #8 , I got the evaluation results. The translational error is 4.029%, the rotational error is 1.317 deg/100m, and ATE is as large as 61.888m.
sequence_09.pdf
Also, here are the relative poses predicted by the network of sequence 09, which contains all poses in one .npy file:
09_bw.zip
BTW, I have used the CUDA 10.2. Will that cause some problem? Thanks for help!

Best

@jytime
Copy link
Owner

jytime commented Feb 4, 2022

Happy the Spring Festival. I would have a look after coming back to the work. I guess I should clean the pose evaluation code and release it later.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants