cs.CVNov 18, 2026

GaussLearner: 3D Gaussian Primitives as Self-Regulating Continual Learners

Authors: Gaoxiang Linghu, Haonan Zhou, Xuanhe Liu, Weili Xu, Kaiyue Zhou, Jun Xiao, Gaoang Wang

Organizations: Zhejiang University · Chengdu Minto Tech

Abstract

Continual learning of 3D Gaussian fields aims to maintain faithful reconstructions of the evolving world. However, directly fine-tuning an existing Gaussian field on new observations can cause catastrophic forgetting of previously reconstructed content. To address this challenge, existing methods employ selective local optimization, replay, or parameter regularization, yet still struggle to balance adaptation to scene changes with preservation of still-valid content. Our key insight is to view 3D Gaussian primitives themselves as self-regulating continual learners. Based on this insight, we introduce GaussLearner, which equips each Gaussian with a closed continual-learning loop connecting observation, self-regulation, and adaptation at the primitive level. At each novel state, GaussLearner first attributes pixel-wise inconsistencies in current observations to their contributing Gaussians. Each Gaussian then self-regulates through a learning weight that balances current observations against cross-state reliability. This weight governs both Gaussian parameter adaptation and optimizer memory refresh. As current inconsistencies are resolved, the adaptation outcome renews each Gaussian's reliability for future states. The resulting field is a globally coupled population of locally self-regulating learners. Across scenes from CL-Splats, WAT, and our CL-3D dataset, GaussLearner improves average PSNR by 3.41 dB over the strongest baseline. On CL-3D, it achieves the best reconstruction quality on both changed and unchanged views.

Figures & tables

Explore similar work

CardsList
  1. AdaptiveSplat:Texture Aware Controllable 3D Gaussian Allocation for Feed-Forward Reconstruction

    Jul 5, 2026Badrinath Singhal, Srihari K G, Sreehari Iyer +23D Gaussian Splatting3D Gaussian Pruning

  2. DecoGS: Adaptive Static-Dynamic Decoupling of 3D Gaussians for Free-Viewpoint Video Streaming

    Sep 15, 2026Idil Sulo, Alexey Supikov, Ilke Demir +13D Gaussian SplattingStreaming 3D Reconstruction

  3. 3D Gaussian Splatting within Self-Learned Neural View-Dependent Color Fields

    Nov 2, 2026Yixiao Yang, Yu-Shen Liu, Zhizhong Han3DGS TrainingNovel View Synthesis