PAGS: Priority-Adaptive Gaussian Splatting for Dynamic Driving Scenes

2025-10-14

AI Summary

Existing dynamic 3D urban scene reconstruction methods suffer from a fundamental trade-off between geometric fidelity and computational efficiency, primarily due to semantic-agnostic, uniform resource allocation. This work proposes a semantics-aware Gaussian lattice rendering framework tailored for autonomous driving, introducing— for the first time—a task-prioritization mechanism that jointly optimizes high-fidelity reconstruction of critical objects (e.g., vehicles, pedestrians) and overall system efficiency. Key innovations include semantics-guided pruning and regularization, depth-prepass culling, hybrid importance scoring, and adaptive Gaussian distribution modeling, collectively forming a priority-driven rendering pipeline. Evaluated on Waymo and KITTI benchmarks, our method achieves significant improvements in reconstruction accuracy for critical objects, reduces training time by over 40%, and enables real-time rendering at >350 FPS.

Technology Category

• Computer Vision: 3D Computer Vision
• Search and Optimization: Sampling/Simulation-based Search
• Planning, Routing, and Scheduling: Replanning and Plan Repair

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Reconstructing dynamic 3D urban scenes is crucial for autonomous driving, yet current methods face a stark trade-off between fidelity and computational cost. This inefficiency stems from their semantically agnostic design, which allocates resources uniformly, treating static backgrounds and safety-critical objects with equal importance. To address this, we introduce Priority-Adaptive Gaussian Splatting (PAGS), a framework that injects task-aware semantic priorities directly into the 3D reconstruction and rendering pipeline. PAGS introduces two core contributions: (1) Semantically-Guided Pruning and Regularization strategy, which employs a hybrid importance metric to aggressively simplify non-critical scene elements while preserving fine-grained details on objects vital for navigation. (2) Priority-Driven Rendering pipeline, which employs a priority-based depth pre-pass to aggressively cull occluded primitives and accelerate the final shading computations. Extensive experiments on the Waymo and KITTI datasets demonstrate that PAGS achieves exceptional reconstruction quality, particularly on safety-critical objects, while significantly reducing training time and boosting rendering speeds to over 350 FPS.

Problem

• Addresses fidelity-computation trade-off in dynamic 3D urban reconstruction
• Allocates resources uniformly without semantic priorities in reconstruction
• Improves reconstruction quality while reducing training time and cost

Innovation

• Priority-adaptive framework with semantic task priorities
• Semantically-guided pruning strategy for non-critical elements
• Priority-driven rendering pipeline accelerating occlusion culling