Search papers, labs, and topics across Lattice.
This paper introduces Semantic Motion Graph (SMG), a novel framework for enhancing dynamic Gaussian splatting from monocular videos by leveraging the structured coherence of semantic motion. By modeling Gaussian motion as low-rank semantic motion and utilizing reliable graph nodes to guide less reliable ones, SMG effectively mitigates issues of overfitting and occlusion in complex scenes. Extensive experiments on a new multiview dataset reveal that SMG achieves state-of-the-art performance in dynamic Gaussian splatting, demonstrating its robustness in challenging real-world scenarios.
Semantic Motion Graphs redefine how we model dynamic scenes, achieving state-of-the-art results by harnessing the power of semantic coherence in motion.
We study dynamic Gaussian Splatting from monocular videos. While recent advancements in dynamic Gaussian splatting offer a promising foundation for modeling dynamic scenes, they often overfit to the training views and fail under occlusion or complex scene motion due to the lack of reliable regularization signals in under-constrained regions. We propose Semantic Motion Graph (SMG), a novel approach models the Gaussian motion as the low-rank semantic motion. Our key insight is that the real-world scene motion is often structured by semantic coherence: regions that are spatially close and semantically related tend to exhibit consistent dynamics. To leverage this prior, we construct SMG to model structured motion of the scene. The Gaussian motion is driven by the motion of SMG nodes. We further observe that the uncertainty of Gaussian motion arises from both unreliable off-the-shelf priors and weakly constrained regions during optimization. SMG addresses this by using reliable graph nodes to guide the motion of nearby unreliable nodes. To evaluate dynamic Gaussian splatting under challenging real-world scenarios, we introduce a new multiview dataset collected under an ego-exo setup. Extensive experiments demonstrate that SMG achieves state-of-the-art performance on monocular dynamic Gaussian splatting across challenging real-world benchmarks. Project page: https://smg-gaussian.github.io/.
