Learning Dynamic Scene Reconstruction with Sinusoidal Geometric Priors

TL;DR

SirenPose introduces a sinusoidal representation-based loss with geometric priors and physics-inspired constraints, significantly enhancing dynamic 3D scene reconstruction accuracy and spatiotemporal consistency in fast-moving, multi-target scenes.

Contribution

The paper presents SirenPose, a novel loss function combining sinusoidal activations and geometric priors, along with an expanded dataset, to improve dynamic scene reconstruction.

Findings

Improved spatiotemporal consistency metrics.

Enhanced accuracy in rapid motion scenes.

Superior performance over prior methods.

Abstract

We propose SirenPose, a novel loss function that combines the periodic activation properties of sinusoidal representation networks with geometric priors derived from keypoint structures to improve the accuracy of dynamic 3D scene reconstruction. Existing approaches often struggle to maintain motion modeling accuracy and spatiotemporal consistency in fast moving and multi target scenes. By introducing physics inspired constraint mechanisms, SirenPose enforces coherent keypoint predictions across both spatial and temporal dimensions. We further expand the training dataset to 600,000 annotated instances to support robust learning. Experimental results demonstrate that models trained with SirenPose achieve significant improvements in spatiotemporal consistency metrics compared to prior methods, showing superior performance in handling rapid motion and complex scene changes.