DualMLP: a two-stream fusion model for 3D point cloud classification

Paul, Sneha; Patterson, Zachary; Bouguila, Nizar

doi:10.1007/s00371-023-03114-3

Cited by 3 publications

(2 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As shown in Table 1, modern feature learning neural networks for point cloud data have been conducted on classification tasks using the ModelNet40 and ScanObjectNN datasets, as well as part segmentation tasks using the ShapeNetPart dataset [1,6,7,[11][12][13][14][15][16][17][18][19][20][21][22][23][24][25][26][27]. Most studies originate from the pioneering work on PointNet by Qi et al [1].…”

Section: Feature Learning On Point Cloudsmentioning

confidence: 99%

“…Recently, several advancements have been made in point cloud feature learning. DualMLP, introduced by Paul et al [24], extends the architecture of PointMLP to address the trade-off between the number of input points and computational cost. It consists of a SparseNet for processing a small subset of points and a DenseNet for handling a larger number of points, effectively balancing computational efficiency and scene understanding while achieving improved performance on the ScanObjectNN dataset compared to PointMLP.…”

Section: Feature Learning On Point Cloudsmentioning

confidence: 99%

See 1 more Smart Citation

Advanced Feature Learning on Point Clouds Using Multi-Resolution Features and Learnable Pooling

Wijaya,

Paek,

Kong

2024

Remote Sensing

View full text Add to dashboard Cite

Existing point cloud feature learning networks often learn high-semantic point features representing the global context by incorporating sampling, neighborhood grouping, neighborhood-wise feature learning, and feature aggregation. However, this process may result in a substantial loss of granular information due to the sampling operation and the widely-used max pooling feature aggregation, which neglects information from non-maximum point features. Consequently, the resulting high-semantic point features could be insufficient to represent the local context, hindering the network’s ability to distinguish fine shapes. To address this problem, we propose PointStack, a novel point cloud feature learning network that utilizes multi-resolution feature learning and learnable pooling (LP). PointStack aggregates point features of various resolutions across multiple layers to capture both high-semantic and high-resolution information. The LP function calculates the weighted sum of multi-resolution point features through an attention mechanism with learnable queries, enabling the extraction of all available information. As a result, PointStack can effectively represent both global and local contexts, allowing the network to comprehend both the global structure and local shape details. PointStack outperforms various existing feature learning networks for shape classification and part segmentation on the ScanObjectNN and ShapeNetPart datasets, achieving 87.2% overall accuracy and instance mIoU.

show abstract