MiLA: Multi-view Intensive-fidelity Long-term Video Generation World Model for Autonomous Driving
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_ | 1866915206391660544 |
|---|---|
| author | Wang, Haiguang Liu, Daqi Xie, Hongwei Liu, Haisong Ma, Enhui Yu, Kaicheng Wang, Limin Wang, Bing |
| author_facet | Wang, Haiguang Liu, Daqi Xie, Hongwei Liu, Haisong Ma, Enhui Yu, Kaicheng Wang, Limin Wang, Bing |
| contents | In recent years, data-driven techniques have greatly advanced autonomous driving systems, but the need for rare and diverse training data remains a challenge, requiring significant investment in equipment and labor. World models, which predict and generate future environmental states, offer a promising solution by synthesizing annotated video data for training. However, existing methods struggle to generate long, consistent videos without accumulating errors, especially in dynamic scenes. To address this, we propose MiLA, a novel framework for generating high-fidelity, long-duration videos up to one minute. MiLA utilizes a Coarse-to-Re(fine) approach to both stabilize video generation and correct distortion of dynamic objects. Additionally, we introduce a Temporal Progressive Denoising Scheduler and Joint Denoising and Correcting Flow modules to improve the quality of generated videos. Extensive experiments on the nuScenes dataset show that MiLA achieves state-of-the-art performance in video generation quality. For more information, visit the project website: https://github.com/xiaomi-mlab/mila.github.io. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2503_15875 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | MiLA: Multi-view Intensive-fidelity Long-term Video Generation World Model for Autonomous Driving Wang, Haiguang Liu, Daqi Xie, Hongwei Liu, Haisong Ma, Enhui Yu, Kaicheng Wang, Limin Wang, Bing Computer Vision and Pattern Recognition In recent years, data-driven techniques have greatly advanced autonomous driving systems, but the need for rare and diverse training data remains a challenge, requiring significant investment in equipment and labor. World models, which predict and generate future environmental states, offer a promising solution by synthesizing annotated video data for training. However, existing methods struggle to generate long, consistent videos without accumulating errors, especially in dynamic scenes. To address this, we propose MiLA, a novel framework for generating high-fidelity, long-duration videos up to one minute. MiLA utilizes a Coarse-to-Re(fine) approach to both stabilize video generation and correct distortion of dynamic objects. Additionally, we introduce a Temporal Progressive Denoising Scheduler and Joint Denoising and Correcting Flow modules to improve the quality of generated videos. Extensive experiments on the nuScenes dataset show that MiLA achieves state-of-the-art performance in video generation quality. For more information, visit the project website: https://github.com/xiaomi-mlab/mila.github.io. |
| title | MiLA: Multi-view Intensive-fidelity Long-term Video Generation World Model for Autonomous Driving |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2503.15875 |