English(EN) FrozenDrive: Zero-Shot Text-Guided Driving Scene Generation and Data Augmentation with Parameter-Free Frozen Diffusion Model

FrozenDrive 使用无参数扩散模型进行合成驾驶场景生成

作者 PulseAugur 编辑部 · [2 个来源] · 2026-06-18 11:34

研究人员开发了 FrozenDrive，一个使用无参数扩散模型生成合成驾驶场景的新框架。该方法通过保留预训练知识和改善文本对齐来解决当前模型的局限性，即使在恶劣天气条件或稀有物体类别下，也能生成一致的多视图和时间连贯的场景。当应用于 nuScenes 数据集时，使用 FrozenDrive 增强的数据显著提高了自动驾驶模型在夜间和雨天等挑战性场景下的性能和鲁棒性。 AI

影响为自动驾驶系统提供更强大、更多样化的训练数据，可能加速其开发和部署。

排序理由该集群描述了一篇详细介绍使用扩散模型进行合成数据生成的新颖方法的最新研究论文。

在 arXiv cs.CV 阅读 →

AI 生成摘要 · Google Gemini · 来自 2 个来源。我们如何撰写摘要 →

报道来源 [2]

arXiv cs.CV TIER_1 English(EN) · Yuhwan Jeong, Hyeonseong Kim, Daehyun We, Seonkyu Song, Jinnyeong Yang, Hyun-Kurl Jang, Youngho Yoon, Kuk-Jin Yoon · 2026-06-19 04:00

FrozenDrive: Zero-Shot Text-Guided Driving Scene Generation and Data Augmentation with Parameter-Free Frozen Diffusion Model

arXiv:2606.20110v1 Announce Type: new Abstract: Synthetic data for autonomous driving is surging, powered by diffusion models that promise scalable scene generation. Yet key obstacles remain, as enforcing multi-view and temporal consistency often relies on backbone fine-tuning or…
arXiv cs.CV TIER_1 English(EN) · Kuk-Jin Yoon · 2026-06-18 11:34

FrozenDrive: Zero-Shot Text-Guided Driving Scene Generation and Data Augmentation with Parameter-Free Frozen Diffusion Model

Synthetic data for autonomous driving is surging, powered by diffusion models that promise scalable scene generation. Yet key obstacles remain, as enforcing multi-view and temporal consistency often relies on backbone fine-tuning or added layers, which erodes pre-trained knowledg…

报道来源 [2]

FrozenDrive: Zero-Shot Text-Guided Driving Scene Generation and Data Augmentation with Parameter-Free Frozen Diffusion Model

FrozenDrive: Zero-Shot Text-Guided Driving Scene Generation and Data Augmentation with Parameter-Free Frozen Diffusion Model

相关实体

相关话题