Contrast with Reconstruct: Contrastive 3D Representation Learning Guided by Generative Pretraining

TL;DR

This paper introduces ReCon, a unified 3D representation learning framework that combines contrastive and generative paradigms through ensemble distillation, achieving state-of-the-art results on 3D classification tasks.

Contribution

ReCon is the first method to effectively unify contrastive and generative 3D learning paradigms using ensemble distillation and a novel encoder-decoder block, overcoming their individual limitations.

Findings

ReCon achieves 91.26% accuracy on ScanObjectNN.

ReCon outperforms previous state-of-the-art methods.

The proposed approach reduces over-fitting and pattern mismatch issues.

Abstract

Mainstream 3D representation learning approaches are built upon contrastive or generative modeling pretext tasks, where great improvements in performance on various downstream tasks have been achieved. However, we find these two paradigms have different characteristics: (i) contrastive models are data-hungry that suffer from a representation over-fitting issue; (ii) generative models have a data filling issue that shows inferior data scaling capacity compared to contrastive models. This motivates us to learn 3D representations by sharing the merits of both paradigms, which is non-trivial due to the pattern difference between the two paradigms. In this paper, we propose Contrast with Reconstruct (ReCon) that unifies these two paradigms. ReCon is trained to learn from both generative modeling teachers and single/cross-modal contrastive teachers through ensemble distillation, where the generative student guides the contrastive student. An encoder-decoder style ReCon-block is proposed that transfers knowledge through cross attention with stop-gradient, which avoids pretraining over-fitting and pattern difference issues. ReCon achieves a new state-of-the-art in 3D representation learning, e.g., 91.26% accuracy on ScanObjectNN. Codes have been released at https://github.com/qizekun/ReCon.