Less is More: Efficient Point Cloud Reconstruction via Multi-Head Decoders

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Alonso, Pedro, Li, Tianrui, Li, Chongshou
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916757183135744
author Alonso, Pedro
Li, Tianrui
Li, Chongshou
author_facet Alonso, Pedro
Li, Tianrui
Li, Chongshou
contents We challenge the common assumption that deeper decoder architectures always yield better performance in point cloud reconstruction. Our analysis reveals that, beyond a certain depth, increasing decoder complexity leads to overfitting and degraded generalization. Additionally, we propose a novel multi-head decoder architecture that exploits the inherent redundancy in point clouds by reconstructing complete shapes from multiple independent heads, each operating on a distinct subset of points. The final output is obtained by concatenating the predictions from all heads, enhancing both diversity and fidelity. Extensive experiments on ModelNet40 and ShapeNetPart demonstrate that our approach achieves consistent improvements across key metrics--including Chamfer Distance (CD), Hausdorff Distance (HD), Earth Mover's Distance (EMD), and F1-score--outperforming standard single-head baselines. Our findings highlight that output diversity and architectural design can be more critical than depth alone for effective and efficient point cloud reconstruction.
format Preprint
id arxiv_https___arxiv_org_abs_2505_19057
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Less is More: Efficient Point Cloud Reconstruction via Multi-Head Decoders
Alonso, Pedro
Li, Tianrui
Li, Chongshou
Computer Vision and Pattern Recognition
We challenge the common assumption that deeper decoder architectures always yield better performance in point cloud reconstruction. Our analysis reveals that, beyond a certain depth, increasing decoder complexity leads to overfitting and degraded generalization. Additionally, we propose a novel multi-head decoder architecture that exploits the inherent redundancy in point clouds by reconstructing complete shapes from multiple independent heads, each operating on a distinct subset of points. The final output is obtained by concatenating the predictions from all heads, enhancing both diversity and fidelity. Extensive experiments on ModelNet40 and ShapeNetPart demonstrate that our approach achieves consistent improvements across key metrics--including Chamfer Distance (CD), Hausdorff Distance (HD), Earth Mover's Distance (EMD), and F1-score--outperforming standard single-head baselines. Our findings highlight that output diversity and architectural design can be more critical than depth alone for effective and efficient point cloud reconstruction.
title Less is More: Efficient Point Cloud Reconstruction via Multi-Head Decoders
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2505.19057