PointNeXt

Bases: NamedTuple

Input features and point cloud of one encoder block, recorded before it downsamples.

Bases: Module

One encoder stage: an optional set-abstraction downsampling followed by depth inverted residual blocks.

Bases: Module

Stack of PointNeXtEncoderBlock stages, each preceded by a set-abstraction downsampling.

When return_intermediates=True is passed to forward, the pre-downsampling features of every stage are returned in coarse-to-fine order, ready to be consumed as decoder skips.

Parameters:

• channels (Sequence[int]) –

  Channels of the stem output followed by the output of each encoder block.

• spatial_dim (int, default: 3 ) –

  Spatial dimension of the input point cloud.

• depths (Sequence[int]) –

  Number of residual blocks in each encoder block.

• expansion (Union[int, Sequence[int]], default: 4 ) –

  Bottleneck expansion factor of the residual blocks.

• ratios (Sequence[float]) –

  Sampling ratio of the set-abstraction preceding each encoder block.

• radiuses (Sequence[Union[float, Sequence[float]]]) –

  Ball-query radius for each channel level.

• num_neighbors (Sequence[Union[int, Sequence[int]]]) –

  Maximum number of neighbors for each channel level.

• sa_layers (int, default: 1 ) –

  Number of MLP layers inside each set-abstraction.

• sa_use_res (bool, default: True ) –

  Whether the set-abstractions use a residual connection.

• act (Union[str, Callable, None], default: 'relu' ) –

  Activation function to use for the encoder blocks.

• act_kwargs (Optional[Dict[str, Any]], default: None ) –

  Keyword arguments for the activation function.

• act_first (bool, default: False ) –

  Whether to apply the activation function before the normalization.

• norm (Union[str, Callable, None], default: 'batch_norm' ) –

  Normalization function to use for the encoder blocks.

• norm_kwargs (Optional[Dict[str, Any]], default: None ) –

  Keyword arguments for the normalization function.

• bias (Union[bool, List[bool]], default: True ) –

  Whether to use a bias for the encoder blocks.

• add_self_loops (bool, default: False ) –

  Whether the neighborhood graphs include self-loops.

• aggr (AggrType, default: 'max' ) –

  Aggregation used to pool neighbor features.

Bases: Module

The PointNeXt decoder, using the Feature Propagation (FP) module from the PointNet++ architecture.

Parameters:

• channels (Sequence[int]) –

  List of channels for each FP block. The first element should correspond to the last channel of the encoder.

• skip_channels (Sequence[int]) –

  List of channels for the skip connections. This is usually the same as the first \(N-1\) channels of the encoder, in reverse order.

• depths (Sequence[int]) –

  List of depths for each FP block.

• spatial_dim (int, default: 3 ) –

  Spatial dimension of the input point cloud.

• dropout (float, default: 0.0 ) –

  Dropout rate before the classification head.

• act (Union[str, Callable, None], default: 'relu' ) –

  Activation function to use for the FP blocks.

• act_kwargs (Optional[Dict[str, Any]], default: None ) –

  Keyword arguments for the activation function.

• act_first (bool, default: False ) –

  Whether to apply the activation function before the normalization.

• norm (Union[str, Callable, None], default: 'batch_norm' ) –

  Normalization function to use for the FP blocks.

• norm_kwargs (Optional[Dict[str, Any]], default: None ) –

  Keyword arguments for the normalization function.

• bias (Union[bool, List[bool]], default: True ) –

  Whether to use a bias for the FP blocks.

Bases: Module

PointNeXt decoder for part segmentation (ShapeNetPart).

Uses the same FP block layout as PointNeXtDecoder, with two global feature convolutions and shape-category conditioning. At the shallowest decoder stage, the skip features are augmented with max-pooled global features from two encoder levels and a shape-category one-hot vector before the FP layer.

This matches the reference PointNextPartDecoder with cls_map='curvenet'.

Parameters:

• channels (Sequence[int]) –

  List of channels for each FP block.

• skip_channels (Sequence[int]) –

  List of channels for the skip connections.

• depths (Sequence[int]) –

  List of depths for each FP block.

• global_conv1_in (int) –

  Input channels for global_conv1 (typically encoder_channels[-2]).

• global_conv2_in (int) –

  Input channels for global_conv2 (typically encoder_channels[-1]).

• num_categories (int, default: 16 ) –

  Number of shape categories (16 for ShapeNetPart).

Bases: SegmentationModel

PointNeXt part segmentation model for ShapeNetPart.

Uses the same encoder as PointNeXtSegmentation but replaces the decoder with PointNeXtPartDecoder (global feature conditioning on shape category) and adds a head with global max+avg pooling.

Parameters:

• in_channels (int) –

  Number of input feature channels.

• num_classes (int) –

  Number of part classes (50 for ShapeNetPart).

• num_categories (int) –

  Number of shape categories (16 for ShapeNetPart).

• stem_channels (Optional[Union[int, Sequence[int]]], default: None ) –

  Stem MLP channel sizes.

• encoder_channels (Sequence[int]) –

  Encoder channel dimensions per stage.

• encoder_depths (Sequence[int]) –

  Residual block depths per encoder stage.

• encoder_expansion (Union[int, Sequence[int]], default: 4 ) –

  InvResMLP expansion ratio.

• sa_layers (int, default: 1 ) –

  Number of SA conv layers per block.

• sa_use_res (bool, default: True ) –

  Whether SA blocks use residual connections.

• decoder_channels (Sequence[int]) –

  Decoder channel dimensions per stage.

• decoder_depths (Sequence[int]) –

  FP block depths per decoder stage.

• ratios (Sequence[float]) –

  FPS downsampling ratios.

• radiuses (Sequence[Union[float, Sequence[float]]]) –

  Ball-query radii.

• num_neighbors (Sequence[Union[int, Sequence[int]]]) –

  Max neighbors per ball query.

Methods:

• configure_stem –

  Build the stem MLP lifting the input features, or None when stem_channels is unset.

• configure_encoder –

  Build the PointNeXtEncoder backbone.

• configure_decoder –

  Build the PointNeXtPartDecoder upsampling the coarsest features back through the encoder skips.

Attributes:

• num_features (int) –

  Channel count \(C\) entering the head: decoder features concatenated with their global max and mean pools.

num_features: int

configure_stem() -> Optional[Module]

configure_encoder() -> PointNeXtEncoder

configure_decoder() -> PointNeXtPartDecoder

Bases: ClassificationModel

PointNeXt classification model as described in the paper PointNeXt: Revisiting PointNet++ with Improved Training and Scaling Strategies by Guocheng Qian, Yuchen Li, Houwen Peng, Jinjie Mai, Hasan Abed Al Kader Hammoud, Mohamed Elhoseiny, Bernard Ghanem.

PointNeXt modernizes PointNet++ through improved training strategies and architectural enhancements, achieving state-of-the-art performance while maintaining efficiency. The model introduces Inverted Residual MLP (InvResMLP) blocks, separable MLPs, and relative position normalization to enable effective model scaling.

Parameters:

• in_channels (int) –

  Number of input channels (typically 3 for XYZ coordinates, or 6 for XYZ + RGB, or more with additional features like normal).

• num_classes (int) –

  Number of output classes for classification.

• stem_channels (Optional[Union[int, Sequence[int]]], default: None ) –

  Number of channels in the stem MLP layer(s) that map input to higher dimension. If None, no stem is used.

• encoder_channels (Sequence[int]) –

  Channel dimensions for each encoder block. The number of channels should be equal to the number of blocks + 1.

• encoder_depths (Sequence[int]) –

  Number of blocks in each encoder stage after the initial SA block.

• encoder_expansion (Union[int, Sequence[int]], default: 4 ) –

  Expansion ratio to determine the hidden channels of the encoder blocks.

• ratios (Sequence[float]) –

  Downsampling sampling ratios for each encoder stage. The number of ratios should be equal to the number of blocks.

• radiuses (Sequence[Union[float, Sequence[float]]]) –

  Query radius for neighborhood grouping in each stage. The number of radiuses should be equal to the number of blocks + 1.

• num_neighbors (Sequence[Union[int, Sequence[int]]]) –

  Maximum number of neighbors for each encoder stage. The number of num_neighbors should be equal to the number of blocks + 1.

• add_self_loops (bool, default: False ) –

  Whether to include the center point as its own neighbor in grouping operations.

• spatial_dim (int, default: 3 ) –

  Spatial dimensionality of point clouds (typically 3).

• act (Union[str, Callable, None], default: 'relu' ) –

  Activation function.

• act_kwargs (Optional[Dict[str, Any]], default: None ) –

  Additional arguments for the activation function.

• act_first (bool, default: False ) –

  Whether to apply activation before normalization.

• norm (Union[str, Callable, None], default: 'batch_norm' ) –

  Normalization layer type.

• norm_kwargs (Optional[Dict[str, Any]], default: None ) –

  Additional arguments for normalization layer.

• bias (Union[bool, List[bool]], default: True ) –

  Whether to use bias in linear / MLP layers.

• dropout (float, default: 0.0 ) –

  Dropout probability before the classification head.

• global_pool (PoolLike, default: 'max' ) –

  Global pooling operation for final feature aggregation.

Methods:

• configure_stem –

  Build the stem MLP lifting the input features, or None when stem_channels is unset.

• configure_encoder –

  Build the PointNeXtEncoder backbone.

• configure_global_sa –

  Build the global set-abstraction applied before pooling, or None when global_sa_channels is unset.

Attributes:

• num_features (int) –

  Feature dimension \(C\) fed to the classification head, after the optional global set-abstraction.

num_features: int

configure_stem() -> Optional[Module]

configure_encoder() -> PointNeXtEncoder

configure_global_sa() -> Optional[PointNeXtSetAbstraction]

Bases: SegmentationModel

PointNeXt segmentation model as described in the paper PointNeXt: Revisiting PointNet++ with Improved Training and Scaling Strategies by Guocheng Qian, Yuchen Li, Houwen Peng, Jinjie Mai, Hasan Abed Al Kader Hammoud, Mohamed Elhoseiny, Bernard Ghanem.

PointNeXt modernizes PointNet++ through improved training strategies and architectural enhancements, achieving state-of-the-art performance while maintaining efficiency. The model introduces Inverted Residual MLP (InvResMLP) blocks, separable MLPs, and relative position normalization to enable effective model scaling.

Parameters:

• in_channels (int) –

  Number of input channels (typically 3 for XYZ coordinates, or 6 for XYZ + RGB, or more with additional features like normal).

• num_classes (int) –

  Number of output classes for segmentation.

• stem_channels (Optional[Union[int, Sequence[int]]], default: None ) –

  Number of channels in the stem MLP layer(s) that map input to higher dimension. If None, no stem is used.

• encoder_channels (Sequence[int]) –

  Channel dimensions for each encoder block. The number of channels should be equal to the number of blocks + 1.

• encoder_depths (Sequence[int]) –

  Number of layers in each encoder block after the initial SA block.

• encoder_expansion (Union[int, Sequence[int]], default: 4 ) –

  Expansion ratio to determine the hidden channels of the encoder blocks.

• decoder_channels (Sequence[int]) –

  Channel dimensions for each decoder block. The number of channels should be equal to the number of decoder blocks + 1.

• decoder_depths (Sequence[int]) –

  Number of layers in each decoder stage.

• ratios (Sequence[float]) –

  Downsampling sampling ratios for each encoder stage. The number of ratios should be equal to the number of blocks.

• radiuses (Sequence[Union[float, Sequence[float]]]) –

  Query radius for neighborhood grouping in each stage. The number of radiuses should be equal to the number of blocks + 1.

• num_neighbors (Sequence[Union[int, Sequence[int]]]) –

  Maximum number of neighbors for each encoder stage. The number of num_neighbors should be equal to the number of blocks + 1.

• add_self_loops (bool, default: False ) –

  Whether to include the center point as its own neighbor in grouping operations.

• spatial_dim (int, default: 3 ) –

  Spatial dimensionality of point clouds (typically 3).

• act (Union[str, Callable, None], default: 'relu' ) –

  Activation function.

• act_kwargs (Optional[Dict[str, Any]], default: None ) –

  Additional arguments for the activation function.

• act_first (bool, default: False ) –

  Whether to apply activation before normalization.

• norm (Union[str, Callable, None], default: 'batch_norm' ) –

  Normalization layer type.

• norm_kwargs (Optional[Dict[str, Any]], default: None ) –

  Additional arguments for normalization layer.

• bias (Union[bool, List[bool]], default: True ) –

  Whether to use bias in linear / MLP layers.

• dropout (float, default: 0.0 ) –

  Dropout probability before the classification head.

Methods:

• configure_stem –

  Build the stem MLP lifting the input features, or None when stem_channels is unset.

• configure_encoder –

  Build the PointNeXtEncoder backbone.

• configure_decoder –

  Build the PointNeXtDecoder upsampling the coarsest features back through the encoder skips.

Attributes:

• num_features (int) –

  Feature dimension \(C\) of the decoder output.

num_features: int

configure_stem() -> Optional[Module]

configure_encoder() -> PointNeXtEncoder

configure_decoder() -> PointNeXtDecoder