Target-Guided Adversarial Point Cloud Transformer Towards Recognition Against Real-world Corruptions

TL;DR

This paper introduces APCT, a transformer-based model for 3D point cloud recognition that enhances robustness against real-world corruptions by using adversarial feature erasing guided by pattern analysis during training.

Contribution

The paper proposes a novel architecture with adversarial feature erasing mechanisms, including an Adversarial Significance Identifier and a Target-guided Promptor, to improve robustness in 3D recognition tasks.

Findings

Achieves state-of-the-art results on multiple corruption benchmarks.

Effectively identifies and integrates diverse object patterns during training.

Enhances global structure capture in point cloud recognition models.

Abstract

Achieving robust 3D perception in the face of corrupted data presents an challenging hurdle within 3D vision research. Contemporary transformer-based point cloud recognition models, albeit advanced, tend to overfit to specific patterns, consequently undermining their robustness against corruption. In this work, we introduce the Target-Guided Adversarial Point Cloud Transformer, termed APCT, a novel architecture designed to augment global structure capture through an adversarial feature erasing mechanism predicated on patterns discerned at each step during training. Specifically, APCT integrates an Adversarial Significance Identifier and a Target-guided Promptor. The Adversarial Significance Identifier, is tasked with discerning token significance by integrating global contextual analysis, utilizing a structural salience index algorithm alongside an auxiliary supervisory mechanism. The Target-guided Promptor, is responsible for accentuating the propensity for token discard within the self-attention mechanism, utilizing the value derived above, consequently directing the model attention towards alternative segments in subsequent stages. By iteratively applying this strategy in multiple steps during training, the network progressively identifies and integrates an expanded array of object-associated patterns. Extensive experiments demonstrate that our method achieves state-of-the-art results on multiple corruption benchmarks.