Gen4D: Synthesizing Humans and Scenes in the Wild
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866909639614922752 |
|---|---|
| author | Bright, Jerrin Wang, Zhibo Chen, Yuhao Rambhatla, Sirisha Zelek, John Clausi, David |
| author_facet | Bright, Jerrin Wang, Zhibo Chen, Yuhao Rambhatla, Sirisha Zelek, John Clausi, David |
| contents | Lack of input data for in-the-wild activities often results in low performance across various computer vision tasks. This challenge is particularly pronounced in uncommon human-centric domains like sports, where real-world data collection is complex and impractical. While synthetic datasets offer a promising alternative, existing approaches typically suffer from limited diversity in human appearance, motion, and scene composition due to their reliance on rigid asset libraries and hand-crafted rendering pipelines. To address this, we introduce Gen4D, a fully automated pipeline for generating diverse and photorealistic 4D human animations. Gen4D integrates expert-driven motion encoding, prompt-guided avatar generation using diffusion-based Gaussian splatting, and human-aware background synthesis to produce highly varied and lifelike human sequences. Based on Gen4D, we present SportPAL, a large-scale synthetic dataset spanning three sports: baseball, icehockey, and soccer. Together, Gen4D and SportPAL provide a scalable foundation for constructing synthetic datasets tailored to in-the-wild human-centric vision tasks, with no need for manual 3D modeling or scene design. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2506_05397 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Gen4D: Synthesizing Humans and Scenes in the Wild Bright, Jerrin Wang, Zhibo Chen, Yuhao Rambhatla, Sirisha Zelek, John Clausi, David Graphics Artificial Intelligence Lack of input data for in-the-wild activities often results in low performance across various computer vision tasks. This challenge is particularly pronounced in uncommon human-centric domains like sports, where real-world data collection is complex and impractical. While synthetic datasets offer a promising alternative, existing approaches typically suffer from limited diversity in human appearance, motion, and scene composition due to their reliance on rigid asset libraries and hand-crafted rendering pipelines. To address this, we introduce Gen4D, a fully automated pipeline for generating diverse and photorealistic 4D human animations. Gen4D integrates expert-driven motion encoding, prompt-guided avatar generation using diffusion-based Gaussian splatting, and human-aware background synthesis to produce highly varied and lifelike human sequences. Based on Gen4D, we present SportPAL, a large-scale synthetic dataset spanning three sports: baseball, icehockey, and soccer. Together, Gen4D and SportPAL provide a scalable foundation for constructing synthetic datasets tailored to in-the-wild human-centric vision tasks, with no need for manual 3D modeling or scene design. |
| title | Gen4D: Synthesizing Humans and Scenes in the Wild |
| topic | Graphics Artificial Intelligence |
| url | https://arxiv.org/abs/2506.05397 |