Effective Gaussian Management for High-fidelity Object Reconstruction

2025-09-16

AI Summary

Traditional Gaussian splatting suffers from attribute assignment ambiguity and gradient conflicts in high-fidelity object reconstruction, limiting reconstruction quality. To address this, we propose a model-agnostic dynamic Gaussian management framework. First, we introduce a surface-reconstruction-supervised dynamic densification strategy for geometry-guided Gaussian distribution optimization. Second, we incorporate adaptive spherical harmonic (SH) order adjustment and gradient-magnitude-based representation optimization to dynamically activate either SH coefficients or surface normals, thereby mitigating gradient conflicts. Third, we design a task-decoupled pruning mechanism enabling efficient model compression under lightweight representations. Experiments demonstrate that our method consistently outperforms state-of-the-art approaches across multiple benchmarks while reducing parameter count by 38% on average—achieving superior reconstruction accuracy without compromising inference efficiency.

Technology Category

• Search and Optimization: Sampling/Simulation-based Search
• Machine Learning: Learning with Manifolds
• Computer Vision: Diffusion Models for Vision

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

This paper proposes an effective Gaussian management approach for high-fidelity object reconstruction. Departing from recent Gaussian Splatting (GS) methods that employ indiscriminate attribute assignment, our approach introduces a novel densification strategy that dynamically activates spherical harmonics (SHs) or normals under the supervision of a surface reconstruction module, which effectively mitigates the gradient conflicts caused by dual supervision and achieves superior reconstruction results. To further improve representation efficiency, we develop a lightweight Gaussian representation that adaptively adjusts the SH orders of each Gaussian based on gradient magnitudes and performs task-decoupled pruning to remove Gaussian with minimal impact on a reconstruction task without sacrificing others, which balances the representational capacity with parameter quantity. Notably, our management approach is model-agnostic and can be seamlessly integrated into other frameworks, enhancing performance while reducing model size. Extensive experiments demonstrate that our approach consistently outperforms state-of-the-art approaches in both reconstruction quality and efficiency, achieving superior performance with significantly fewer parameters.

Problem

• Dynamic activation of SHs and normals to mitigate gradient conflicts
• Adaptive SH order adjustment and task-decoupled pruning for efficiency
• Model-agnostic Gaussian management for improved reconstruction quality

Innovation

• Dynamic spherical harmonics activation strategy
• Adaptive SH order adjustment via gradients
• Task-decoupled pruning for parameter reduction