Lune

CVPR2025顶会

SuperPC: A Single Diffusion Model for Point Cloud Completion, Upsampling, Denoising, and Colorization

Yi Du, Zhipeng Zhao, Shaoshu Su, Sharath Golluri, Haoze Zheng, Runmao Yao, Chen Wang

摘要

Point cloud (PC) processing tasks-such as completion, upsampling, denoising, and colorization-are crucial in applications like autonomous driving and 3D reconstruction. Despite substantial advancements, prior approaches often address each of these tasks independently, with separate models focused on individual issues. However, this isolated approach fails to account for the fact that defects like incompleteness, low resolution, noise, and lack of color frequently coexist, with each defect influencing and correlating with the others. Simply applying these models sequentially can lead to error accumulation from each model, along with increased computational costs. To address these challenges, we introduce SuperPC, the first unified diffusion model capable of concurrently handling all four tasks. Our approach employs a three-level-conditioned diffusion framework, enhanced by a novel spatial-mix-fusion strategy, to leverage the correlations among these four defects for simultaneous, efficient processing. We show that Su-perPC outperforms the state-of-the-art specialized models as well as their combination on all four individual tasks. Project website: https://sairlab.org/superpc/ .

问 Lune

几分钟读透这篇 Paper,不用再花一下午。

Lune 能读完全文,也能读引用它的顶会 Paper,讲清它的结论、证据和局限,每一点都注明出自哪篇 Paper。不用硬啃,也不靠猜。

已有 7 篇顶会 Paper 引用了它。别等审稿人问起,先弄清它们发现了什么。

可以从这些问题问起

Lune 可以查阅

全文 · 引用它的顶会 Paper · 相关的 Paper

获取答案

免费开始,无需绑卡

脉络

  • High-Resolution Image Synthesis with Latent Diffusion Models 引用了 Zero-Shot Text-to-Image Generation
  • High-Resolution Image Synthesis with Latent Diffusion Models 引用了 Denoising Diffusion Probabilistic Models
  • PUFM: Efficient Point Cloud Upsampling via Flow Matching 引用了 Denoising Diffusion Probabilistic Models
  • Learning Latent Proxies for Controllable Single-Image Relighting 引用了 High-Resolution Image Synthesis with Latent Diffusion Models
  • Learning Latent Proxies for Controllable Single-Image Relighting 引用了 Denoising Diffusion Probabilistic Models
  • PQDT: Pseudo-Query Dual Transformer for Robust Point Cloud Restoration 引用了 Denoising Diffusion Probabilistic Models
  • RePaint: Inpainting using Denoising Diffusion Probabilistic Models 引用了 Denoising Diffusion Probabilistic Models
它引用的引用它的从左到右按年份排列,引用越多,位置越高。

引用它的顶会 Paper7

  1. MoRe: Motion-aware Feed-forward 4D Reconstruction TransformerJuntong Fang, Zequn Chen, Weiqi Zhang, Donglin Di 等CVPR 2026 · 8 次引用
  2. Learning Latent Proxies for Controllable Single-Image RelightingHaoze Zheng, Zihao Wang, Xianfeng Wu, Yajing Bai 等CVPR 2026 · 1 次引用
  3. PUFM: Efficient Point Cloud Upsampling via Flow MatchingZhi-Song Liu, Chenhang He, Yakun Ju, Lei LiAAAI 2026
  4. LiNeXt: Revisiting LiDAR Completion with Efficient Non-Diffusion ArchitecturesWenzhe He, Xiaojun Chen, Ruiqi Wang, Ruihui Li 等AAAI 2026
  5. PQDT: Pseudo-Query Dual Transformer for Robust Point Cloud RestorationHaoqing Wu, Alexa Nawotki, Jochen GarckeCVPR 2026

它引用的 Paper26

  1. Denoising Diffusion Probabilistic ModelsJonathan Ho, Ajay Jain, Pieter AbbeelNeurIPS 2020 · 35,902 次引用
  2. An Image is Worth 16x16 Words: Transformers for Image Recognition at ScaleAlexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn 等ICLR 2021 · 21,477 次引用
  3. High-Resolution Image Synthesis with Latent Diffusion ModelsRobin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser 等CVPR 2022 · 13,123 次引用
  4. Zero-Shot Text-to-Image GenerationAditya Ramesh, Mikhail Pavlov, Gabriel Goh, Scott Gray 等ICML 2021 · 6,356 次引用
  5. RePaint: Inpainting using Denoising Diffusion Probabilistic ModelsAndreas Lugmayr, Martin Danelljan, Andrés Romero, Fisher Yu 等CVPR 2022 · 1,425 次引用

相关 Paper

  1. 3D Shape Generation and Completion through Point-Voxel DiffusionLinqi Zhou, Yilun Du, Jiajun WuICCV 2021 · 681 次引用
  2. Diffusion Probabilistic Models for 3D Point Cloud GenerationShitong Luo, Wei HuCVPR 2021
  3. DeepPCD: Enabling AutoCompletion of Indoor Point Clouds with Deep LearningPingping Cai, Sanjib SurUbiComp 2022 · 2 次引用
  4. SPU-IMR: Self-supervised Arbitrary-scale Point Cloud Upsampling via Iterative Mask-recovery NetworkZiming Nie, Qiao Wu, Chenlei Lv, Siwen Quan 等AAAI 2025 · 2 次引用
  5. ComPC: Completing a 3D Point Cloud with 2D Diffusion PriorsTianxin Huang, Zhiwen Yan, Yuyang Zhao, Gim Hee LeeICLR 2025
黄昏的海面,两侧是细线勾勒的悬崖