1 rri_metrics.metrics

rri_metrics.metrics

Low-level RRI metric primitives (point-mesh distances).

This module wraps PyTorch3D geometric distance implementations so downstream code can compute accuracy, completeness, and bidirectional Chamfer distances in a uniform, torch-first manner. Only the GPU-accelerated PyTorch3D path is supported; CPU fallbacks were removed to keep the hot path simple and consistent. The functions remain thin: callers must supply pre-sampled point clouds and mesh tensors. Directional components are returned separately to expose the accuracy/completeness split described in surface_metrics.qmd.

For a point set \(P\) and mesh \(M\), accuracy is point-to-mesh error and completeness is mesh-to-point error. The current scalar point-mesh error is their sum, which is the value consumed by OracleRRI.

1.1 Functions

Name	Description
chamfer_point_mesh	Compute accuracy, completeness, and bidirectional Chamfer for P<->M.
chamfer_point_mesh_batched	Chamfer-like point↔︎mesh distance per example (fully vectorised).

# rri_metrics.metrics { #aria_nbv.rri_metrics.metrics }

`rri_metrics.metrics`

Low-level RRI metric primitives (point-mesh distances).

This module wraps PyTorch3D geometric distance implementations so downstream
code can compute accuracy, completeness, and bidirectional Chamfer distances in
a uniform, torch-first manner. Only the GPU-accelerated PyTorch3D path is
supported; CPU fallbacks were removed to keep the hot path simple and
consistent. The functions remain thin: callers must supply pre-sampled point
clouds and mesh tensors. Directional components are returned separately to
expose the accuracy/completeness split described in ``surface_metrics.qmd``.

For a point set $P$ and mesh $M$, `accuracy` is point-to-mesh error and
`completeness` is mesh-to-point error. The current scalar point-mesh error is
their sum, which is the value consumed by `OracleRRI`.

## Functions

| Name | Description |
| --- | --- |
| [chamfer_point_mesh](aria_nbv.rri_metrics.metrics.chamfer_point_mesh.qmd#aria_nbv.rri_metrics.metrics.chamfer_point_mesh) | Compute accuracy, completeness, and bidirectional Chamfer for P<->M. |
| [chamfer_point_mesh_batched](aria_nbv.rri_metrics.metrics.chamfer_point_mesh_batched.qmd#aria_nbv.rri_metrics.metrics.chamfer_point_mesh_batched) | Chamfer-like point↔mesh distance per example (fully vectorised). |