GaussianVideo: Efficient Video Representation via Hierarchical Gaussian Splatting
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866916558245199872 |
|---|---|
| author | Bond, Andrew Wang, Jui-Hsien Mai, Long Erdem, Erkut Erdem, Aykut |
| author_facet | Bond, Andrew Wang, Jui-Hsien Mai, Long Erdem, Erkut Erdem, Aykut |
| contents | Efficient neural representations for dynamic video scenes are critical for applications ranging from video compression to interactive simulations. Yet, existing methods often face challenges related to high memory usage, lengthy training times, and temporal consistency. To address these issues, we introduce a novel neural video representation that combines 3D Gaussian splatting with continuous camera motion modeling. By leveraging Neural ODEs, our approach learns smooth camera trajectories while maintaining an explicit 3D scene representation through Gaussians. Additionally, we introduce a spatiotemporal hierarchical learning strategy, progressively refining spatial and temporal features to enhance reconstruction quality and accelerate convergence. This memory-efficient approach achieves high-quality rendering at impressive speeds. Experimental results show that our hierarchical learning, combined with robust camera motion modeling, captures complex dynamic scenes with strong temporal consistency, achieving state-of-the-art performance across diverse video datasets in both high- and low-motion scenarios. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2501_04782 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | GaussianVideo: Efficient Video Representation via Hierarchical Gaussian Splatting Bond, Andrew Wang, Jui-Hsien Mai, Long Erdem, Erkut Erdem, Aykut Computer Vision and Pattern Recognition Efficient neural representations for dynamic video scenes are critical for applications ranging from video compression to interactive simulations. Yet, existing methods often face challenges related to high memory usage, lengthy training times, and temporal consistency. To address these issues, we introduce a novel neural video representation that combines 3D Gaussian splatting with continuous camera motion modeling. By leveraging Neural ODEs, our approach learns smooth camera trajectories while maintaining an explicit 3D scene representation through Gaussians. Additionally, we introduce a spatiotemporal hierarchical learning strategy, progressively refining spatial and temporal features to enhance reconstruction quality and accelerate convergence. This memory-efficient approach achieves high-quality rendering at impressive speeds. Experimental results show that our hierarchical learning, combined with robust camera motion modeling, captures complex dynamic scenes with strong temporal consistency, achieving state-of-the-art performance across diverse video datasets in both high- and low-motion scenarios. |
| title | GaussianVideo: Efficient Video Representation via Hierarchical Gaussian Splatting |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2501.04782 |