FatNet: A Feature-attentive Network for 3D Point Cloud Processing

TL;DR

FatNet introduces a feature-attentive neural network layer that combines global and local features, uses dual pooling attention mechanisms, and employs residual feature reuse, achieving state-of-the-art results in 3D point cloud classification and segmentation.

Contribution

The paper proposes a novel FAT layer and a dual pooling attention mechanism, enhancing feature embedding and training efficiency for 3D point cloud analysis.

Findings

Achieves state-of-the-art accuracy on ModelNet40.

Performs competitively on ShapeNet part segmentation.

Demonstrates effective feature embedding improvements.

Abstract

The application of deep learning to 3D point clouds is challenging due to its lack of order. Inspired by the point embeddings of PointNet and the edge embeddings of DGCNNs, we propose three improvements to the task of point cloud analysis. First, we introduce a novel feature-attentive neural network layer, a FAT layer, that combines both global point-based features and local edge-based features in order to generate better embeddings. Second, we find that applying the same attention mechanism across two different forms of feature map aggregation, max pooling and average pooling, gives better performance than either alone. Third, we observe that residual feature reuse in this setting propagates information more effectively between the layers, and makes the network easier to train. Our architecture achieves state-of-the-art results on the task of point cloud classification, as demonstrated on the ModelNet40 dataset, and an extremely competitive performance on the ShapeNet part segmentation challenge.