SimWorld: A Unified Benchmark for Simulator-Conditioned Scene Generation via World Model
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866916811399757824 |
|---|---|
| author | Li, Xinqing Song, Ruiqi Xie, Qingyu Wu, Ye Zeng, Nanxin Ai, Yunfeng |
| author_facet | Li, Xinqing Song, Ruiqi Xie, Qingyu Wu, Ye Zeng, Nanxin Ai, Yunfeng |
| contents | With the rapid advancement of autonomous driving technology, a lack of data has become a major obstacle to enhancing perception model accuracy. Researchers are now exploring controllable data generation using world models to diversify datasets. However, previous work has been limited to studying image generation quality on specific public datasets. There is still relatively little research on how to build data generation engines for real-world application scenes to achieve large-scale data generation for challenging scenes. In this paper, a simulator-conditioned scene generation engine based on world model is proposed. By constructing a simulation system consistent with real-world scenes, simulation data and labels, which serve as the conditions for data generation in the world model, for any scenes can be collected. It is a novel data generation pipeline by combining the powerful scene simulation capabilities of the simulation engine with the robust data generation capabilities of the world model. In addition, a benchmark with proportionally constructed virtual and real data, is provided for exploring the capabilities of world models in real-world scenes. Quantitative results show that these generated images significantly improve downstream perception models performance. Finally, we explored the generative performance of the world model in urban autonomous driving scenarios. All the data and code will be available at https://github.com/Li-Zn-H/SimWorld. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2503_13952 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | SimWorld: A Unified Benchmark for Simulator-Conditioned Scene Generation via World Model Li, Xinqing Song, Ruiqi Xie, Qingyu Wu, Ye Zeng, Nanxin Ai, Yunfeng Computer Vision and Pattern Recognition I.4.8; I.2.10 With the rapid advancement of autonomous driving technology, a lack of data has become a major obstacle to enhancing perception model accuracy. Researchers are now exploring controllable data generation using world models to diversify datasets. However, previous work has been limited to studying image generation quality on specific public datasets. There is still relatively little research on how to build data generation engines for real-world application scenes to achieve large-scale data generation for challenging scenes. In this paper, a simulator-conditioned scene generation engine based on world model is proposed. By constructing a simulation system consistent with real-world scenes, simulation data and labels, which serve as the conditions for data generation in the world model, for any scenes can be collected. It is a novel data generation pipeline by combining the powerful scene simulation capabilities of the simulation engine with the robust data generation capabilities of the world model. In addition, a benchmark with proportionally constructed virtual and real data, is provided for exploring the capabilities of world models in real-world scenes. Quantitative results show that these generated images significantly improve downstream perception models performance. Finally, we explored the generative performance of the world model in urban autonomous driving scenarios. All the data and code will be available at https://github.com/Li-Zn-H/SimWorld. |
| title | SimWorld: A Unified Benchmark for Simulator-Conditioned Scene Generation via World Model |
| topic | Computer Vision and Pattern Recognition I.4.8; I.2.10 |
| url | https://arxiv.org/abs/2503.13952 |