MGPC: Multimodal Network for Generalizable Point Cloud Completion With Modality Dropout and Progressive Decoding

TL;DR

MGPC is a multimodal framework for point cloud completion that combines point clouds, images, and text, using dropout, Transformer fusion, and progressive decoding to enhance robustness and generalization in real-world scenarios.

Contribution

The paper introduces MGPC, a novel multimodal point cloud completion model with modality dropout, Transformer fusion, and progressive decoding, plus a large-scale benchmark MGPC-1M.

Findings

Outperforms prior methods on MGPC-1M and in-the-wild data.

Demonstrates strong generalization to real-world scenarios.

Shows robustness and scalability in diverse conditions.

Abstract

Point cloud completion aims to recover complete 3D geometry from partial observations caused by limited viewpoints and occlusions. Existing learning-based works, including 3D Convolutional Neural Network (CNN)-based, point-based, and Transformer-based methods, have achieved strong performance on synthetic benchmarks. However, due to the limitations of modality, scalability, and generative capacity, their generalization to novel objects and real-world scenarios remains challenging. In this paper, we propose MGPC, a generalizable multimodal point cloud completion framework that integrates point clouds, RGB images, and text within a unified architecture. MGPC introduces an innovative modality dropout strategy, a Transformer-based fusion module, and a novel progressive generator to improve robustness, scalability, and geometric modeling capability. We further develop an automatic data generation pipeline and construct MGPC-1M, a large-scale benchmark with over 1,000 categories and one million training pairs. Extensive experiments on MGPC-1M and in-the-wild data demonstrate that the proposed method consistently outperforms prior baselines and exhibits strong generalization under real-world conditions.