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

Got wrong result using my own data #3

Open
liangsongLee opened this issue Mar 3, 2022 · 4 comments
Open

Got wrong result using my own data #3

liangsongLee opened this issue Mar 3, 2022 · 4 comments

Comments

@liangsongLee
Copy link

Thanks for your great work!
But I got the wrong result while using my own data. I used the surround view four fisheye cameras on my vehicle. Maybe my calibration file is not right. What is the meaning of T_imu_cam in the calibration file? Does the first one have to be an identity matrix?

@liangsongLee liangsongLee changed the title wrong result Got wrong result using my own data Mar 3, 2022
@liangsongLee
Copy link
Author

Is there any requirement for the order of extrinsics parameters in the calibration jason file?

@ameuleman
Copy link
Collaborator

ameuleman commented Mar 3, 2022

Thank you for your interest in our work.
Your setup seems quite different from ours. To accomodate for the larger baseline, I would advise changing the depth range (espetially the minimum depth) and possibly increase the number of depth candidates.
T_imu_cam is a name given by basalt calibration. It has the camera poses (translation and rotation).
The software should work when the first pose is not identity.
The order of parameters (both intrinsic and extrinsic) should be the same as the camera order in the folder where input images are stored. Also, the reference cameras should be set in a way that allows for the entire field of view to be covered (setting all cameras as reference shouldn't be an issue).

@liangsongLee
Copy link
Author

Thank you for your interest in our work. Your setup seems quite different from ours. To accomodate for the larger baseline, I would advise changing the depth range (espetially the minimum depth) and possibly increase the number of depth candidates. T_imu_cam is a name given by basalt calibration. It has the camera poses (translation and rotation). The software should work when the first pose is not identity. The order of parameters (both intrinsic and extrinsic) should be the same as the camera order in the folder where input images are stored. Also, the reference cameras should be set in a way that allows for the entire field of view to be covered (setting all cameras as reference shouldn't be an issue).
Surround view camera
My fisheye cameras are mounted on car, as is shown in the pic, and the fov of them is 206 degrees,all of them have a downward pitch. Can this configuration run with your code, please?

@ameuleman
Copy link
Collaborator

It should work as long as parameters are set correctly (see previous comment, most important being minimum distance candidate that should probably be set to a few meters), although distance estimation will not be accurate in areas seen by only one camera, if this hapens with your layout. Another important step is to define the masks appropriately so that the car parts are masked out when estimating distance.

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