CuriGS: Curriculum-Guided Gaussian Splatting for Sparse View Synthesis

2025-11-19

AI Summary

To address overfitting and geometric distortion in 3D Gaussian Splatting (3DGS) under sparse-view settings—caused by insufficient supervision signals and limited angular coverage—this paper proposes CuriGS, a curriculum-guided reconstruction framework. Methodologically, it introduces a student-teacher collaborative pseudo-view generation mechanism, integrating depth-aware correlation modeling and co-regularization constraints. Furthermore, it establishes a dynamic curriculum learning strategy that jointly evaluates candidate pseudo-views using multi-modal signals—SSIM, LPIPS, and geometric consistency—and adaptively selects high-fidelity views for staged injection into training. This design significantly enhances representational robustness and geometric fidelity under sparse input conditions. Extensive experiments on both synthetic and real-world scenes demonstrate that CuriGS consistently outperforms state-of-the-art methods in rendering quality (PSNR, SSIM, LPIPS) and geometric reconstruction consistency.

Technology Category

• Computer Vision: 3D Computer Vision
• Machine Learning: Multi-instance/Multi-view Learning
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs

Abstract

3D Gaussian Splatting (3DGS) has recently emerged as an efficient, high-fidelity representation for real-time scene reconstruction and rendering. However, extending 3DGS to sparse-view settings remains challenging because of supervision scarcity and overfitting caused by limited viewpoint coverage. In this paper, we present CuriGS, a curriculum-guided framework for sparse-view 3D reconstruction using 3DGS. CuriGS addresses the core challenge of sparse-view synthesis by introducing student views: pseudo-views sampled around ground-truth poses (teacher). For each teacher, we generate multiple groups of student views with different perturbation levels. During training, we follow a curriculum schedule that gradually unlocks higher perturbation level, randomly sampling candidate students from the active level to assist training. Each sampled student is regularized via depth-correlation and co-regularization, and evaluated using a multi-signal metric that combines SSIM, LPIPS, and an image-quality measure. For every teacher and perturbation level, we periodically retain the best-performing students and promote those that satisfy a predefined quality threshold to the training set, resulting in a stable augmentation of sparse training views. Experimental results show that CuriGS outperforms state-of-the-art baselines in both rendering fidelity and geometric consistency across various synthetic and real sparse-view scenes. Project page: https://zijian1026.github.io/CuriGS/

Problem

• Extending 3D Gaussian Splatting to sparse-view settings with limited supervision
• Addressing overfitting caused by limited viewpoint coverage in 3D reconstruction
• Improving rendering fidelity and geometric consistency in sparse-view synthesis

Innovation

• Curriculum-guided Gaussian splatting for sparse view synthesis
• Student views with perturbation levels for training regularization
• Multi-signal metric evaluation for stable view augmentation