You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Add a feature to visualize reconstruction errors on the feature scalar value computed between the feature predicted from the shape model and the original feature value. This would allow users to assess the impact of using "features" during optimization and tune the feature weight accordingly.
Similarly, add quantitative evaluation metrics (compactness, generalization, specificity) on the feature space and shape + feature space.
The text was updated successfully, but these errors were encountered:
Here is the equation for chamfer distance between two point clouds S1 and S2:
It is the L2 distance between each point and its nearest neighbor in the other point cloud, so the main functionality you need nearest neighbors.
I've been using this PyTorch implementation: https://pytorch3d.readthedocs.io/en/latest/modules/loss.html
There are some other repos that implement the functionality in C++ and adapt it for pytorch, for example: https://github.com/otaheri/chamfer_distance
akenmorris
changed the title
Evaulation metrics for feature scalar (if used in optimization)
Evalation metrics for scalars and shape+scalar
May 5, 2024
Feature Summary
The text was updated successfully, but these errors were encountered: