Towards Compact 3D Representations via Point Feature Enhancement Masked Autoencoders

TL;DR

This paper introduces Point-FEMAE, a novel point cloud autoencoder that enhances 3D feature compactness and efficiency by combining global and local branches with a shared Transformer encoder, outperforming existing methods.

Contribution

It proposes a simple yet effective point feature enhancement autoencoder with a local module and shared encoder, improving 3D representation compactness and pre-training efficiency.

Findings

Outperforms baseline Point-MAE by over 5% on ScanObjectNN variants.

Significantly improves pre-training efficiency compared to cross-modal methods.

Achieves state-of-the-art results in downstream 3D recognition tasks.

Abstract

Learning 3D representation plays a critical role in masked autoencoder (MAE) based pre-training methods for point cloud, including single-modal and cross-modal based MAE. Specifically, although cross-modal MAE methods learn strong 3D representations via the auxiliary of other modal knowledge, they often suffer from heavy computational burdens and heavily rely on massive cross-modal data pairs that are often unavailable, which hinders their applications in practice. Instead, single-modal methods with solely point clouds as input are preferred in real applications due to their simplicity and efficiency. However, such methods easily suffer from limited 3D representations with global random mask input. To learn compact 3D representations, we propose a simple yet effective Point Feature Enhancement Masked Autoencoders (Point-FEMAE), which mainly consists of a global branch and a local branch to capture latent semantic features. Specifically, to learn more compact features, a share-parameter Transformer encoder is introduced to extract point features from the global and local unmasked patches obtained by global random and local block mask strategies, followed by a specific decoder to reconstruct. Meanwhile, to further enhance features in the local branch, we propose a Local Enhancement Module with local patch convolution to perceive fine-grained local context at larger scales. Our method significantly improves the pre-training efficiency compared to cross-modal alternatives, and extensive downstream experiments underscore the state-of-the-art effectiveness, particularly outperforming our baseline (Point-MAE) by 5.16%, 5.00%, and 5.04% in three variants of ScanObjectNN, respectively. The code is available at https://github.com/zyh16143998882/AAAI24-PointFEMAE.