RonaldSun / VI-Stereo-DSO

Direct sparse odometry combined with stereo cameras and IMU
387 stars 124 forks source link

Must I provide the ground truth? #13

Open wangyuanbiubiubiu opened 5 years ago

wangyuanbiubiubiu commented 5 years ago

I think this code must provide the ground truth information, may I ask how can I run without providing this information.

ACFFF commented 5 years ago

When I was running my own data, I also encountered the same problem.

wangyuanbiubiubiu commented 5 years ago

When I was running my own data, I also encountered the same problem.

Yes, I found that the code aligns the position of the first frame with the ground truth, so I don't know how to remove it.

RonaldSun commented 5 years ago

I‘ve fixed the problem

shibowing commented 5 years ago

Hi, Thanks for your sharing the code. I have tested the current code with commit "fix bug when no ground truth". However, I found the latest code has a bug in this part (src/FullSystem/FullSystem.cpp). if(gt_path.size() > 0) T_WR_align = T_wc * T_BC.inverse()*gt_pose[index2].inverse();<------especially have bug on this sentence. else T_WR_align = SE3();

Have you tested this version of code on the euroc dataset? I can run code well on the previous version. However, the current version will have a segmenat fault.
Thanks again for sharing the code!

shibowing commented 5 years ago

Hi, Thanks for your sharing the code. I have tested the current code with commit "fix bug when no ground truth". However, I found the latest code has a bug in this part (src/FullSystem/FullSystem.cpp). if(gt_path.size() > 0) T_WR_align = T_wc * T_BC.inverse()*gt_pose[index2].inverse();<------especially have bug on this sentence. else T_WR_align = SE3();

Have you tested this version of code on the euroc dataset? I can run code well on the previous version. However, the current version will have a segmenat fault. Thanks again for sharing the code!

My bad. I have solved the problem. Thank you!