GeoFormer: Learning Point Cloud Completion with Tri-Plane Integrated Transformer

TL;DR

GeoFormer is a novel point cloud completion method that integrates multi-view consistent features and multi-scale upsampling to improve global structure and local detail recovery, achieving state-of-the-art results.

Contribution

It introduces a CCM Feature Enhanced Point Generator and a Multi-scale Geometry-aware Upsampler for improved point cloud completion.

Findings

Outperforms recent methods on PCN, ShapeNet, and KITTI benchmarks.

Achieves state-of-the-art performance in point cloud completion.

Effectively enhances global geometry and local details.

Abstract

Point cloud completion aims to recover accurate global geometry and preserve fine-grained local details from partial point clouds. Conventional methods typically predict unseen points directly from 3D point cloud coordinates or use self-projected multi-view depth maps to ease this task. However, these gray-scale depth maps cannot reach multi-view consistency, consequently restricting the performance. In this paper, we introduce a GeoFormer that simultaneously enhances the global geometric structure of the points and improves the local details. Specifically, we design a CCM Feature Enhanced Point Generator to integrate image features from multi-view consistent canonical coordinate maps (CCMs) and align them with pure point features, thereby enhancing the global geometry feature. Additionally, we employ the Multi-scale Geometry-aware Upsampler module to progressively enhance local details. This is achieved through cross attention between the multi-scale features extracted from the partial input and the features derived from previously estimated points. Extensive experiments on the PCN, ShapeNet-55/34, and KITTI benchmarks demonstrate that our GeoFormer outperforms recent methods, achieving the state-of-the-art performance. Our code is available at \href{https://github.com/Jinpeng-Yu/GeoFormer}{https://github.com/Jinpeng-Yu/GeoFormer}.