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

What is the accuracy of the 3D contour image reconstructed from two or abover 2D images? #75

Open
caibf opened this issue Apr 11, 2024 · 3 comments

Comments

@caibf
Copy link

caibf commented Apr 11, 2024

According to your publicly available paper, the average imaging accuracy is 2.7mm in the absence of specific camera parameters. What is the accuracy we can achieve if we use a high-resolution 2D camera to perform 3D reconstruction within a straight distance of about 15cm?
1712820867267

@vincent-leroy
Copy link
Contributor

I guess it should be similar to the performance of the DTU dataset, since the setup would be roughly equivalent. This really depends on the observation setup, so you would need to quantify the quality in your particular setup. Keep in mind that without known GT cameras, the reconstruction is done up to an unknown scale factor.

@caibf
Copy link
Author

caibf commented Apr 11, 2024

I guess it should be similar to the performance of the DTU dataset, since the setup would be roughly equivalent. This really depends on the observation setup, so you would need to quantify the quality in your particular setup. Keep in mind that without known GT cameras, the reconstruction is done up to an unknown scale factor.

Thank you for your valuable suggestion.
I haven't practiced this open source library yet. Would it be convenient for you to recommend relevant hardware configurations or models for me to refer to? For example, 2D camera configuration or related parameters.

@vincent-leroy
Copy link
Contributor

You can try the model with a simple phone camera and see it's strengths and weaknesses. In our experience you will get the best quantitative results with minimal distortion lenses (model was trained on undistorted images). As for the camera setup, this is really application dependent so I can't tell you beforehand. Luckily it is fast to iterate over setups if you have only a few images.

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