1 vin.modules.scene_field

vin.modules.scene_field

Trainable scene-field projection modules for VIN scorers.

Scene-field construction is deterministic and lives in aria_nbv.vin.scorer_context.build_vin_scorer_scene_field. This module owns the small learnable projection that maps those ordered scalar channels into the voxel feature width consumed by pose-conditioned pooling.

1.1 Classes

Name Description
SceneFieldProjection Project scalar voxel evidence channels into VIN field features.
SceneFieldProjectionConfig Configure the trainable VIN scene-field projection.

1.1.1 SceneFieldProjection

vin.modules.scene_field.SceneFieldProjection(config)

Project scalar voxel evidence channels into VIN field features.

The class subclasses torch.nn.Sequential so models can assign it directly to field_proj while preserving historical checkpoint keys such as field_proj.0.weight, field_proj.1.weight, and field_proj.1.bias.

1.1.2 SceneFieldProjectionConfig

vin.modules.scene_field.SceneFieldProjectionConfig()

Configure the trainable VIN scene-field projection.

1.1.2.1 Attributes

Name Description
target_type Factory target for BaseConfig.setup_target.
in_channels Number of ordered scalar scene-field channels.
field_dim Projected voxel feature dimension.
field_gn_groups Requested GroupNorm groups, clamped to a divisor of field_dim.
Back to top