new E decomposition without svd

[weitong8591]

Changes

As SVD used in E decomposition is not stable for training, we would like to intergrate another methodthat recovers the two solutions for the baseline and the orientation from a given essential matrix using elementary matrix operations. This work was done by Berthold K.P. Horn in 1990, and used in Generalized Differentiable RANSAC to build the pose erorr loss.

Fixes # (issue)

Type of change

• 📚 Documentation Update
• 🧪 Tests Cases
• 🐞 Bug fix (non-breaking change which fixes an issue)
• 🔬 New feature (non-breaking change which adds functionality)
• 🚨 Breaking change (fix or feature that would cause existing functionality to not work as expected)
• 📝 This change requires a documentation update

Checklist

• My code follows the style guidelines of this project
• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• Did you update CHANGELOG in case of a major change?

[weitong8591]

Hi! any suggestions about what to return if there is no solution? It can't pass the tests. link

[johnnv1]

Hi! any suggestions about what to return if there is no solution? It can't pass the tests. link

since we are randomly generating the input, we cannot determine when it'll be able to solve or not... i may say the best case here is to mark the test as xfail -- https://docs.pytest.org/en/stable/how-to/skipping.html#raises-parameter

[weitong8591]

Hi! any suggestions about what to return if there is no solution? It can't pass the tests. link

since we are randomly generating the input, we cannot determine when it'll be able to solve or not... i may say the best case here is to mark the test as xfail -- https://docs.pytest.org/en/stable/how-to/skipping.html#raises-parameter

Thanks! let me try it