研究人员推出了InFlux++，这是一个新的数据集和基准，旨在改进从RGB图像估计动态相机内参。这一进展解决了当前3D计算机视觉算法的局限性，这些算法通常假设相机内参固定，而这在现实世界的视频中并不成立。InFlux++包含一个大型合成数据集，拥有超过441,000帧，以及一个扩展的真实世界基准，拥有超过514,000帧，提供了更多样化的场景、相机运动以及变焦和对焦等内参变化。 AI
影响 通过改进动态相机内参估计，增强了视频3D重建的鲁棒性。
排序理由 该集群描述了一个计算机视觉研究问题的新数据集和基准。
AI 生成摘要 · Google Gemini · 来自 3 个来源。 我们如何撰写摘要 →
Camera intrinsics are vital for recovering 3D structure from 2D video. However, most 3D algorithms assume fixed intrinsics throughout a video, an assumption that often fails for real-world in-the-wild videos. Consequently, estimating per-frame intrinsics from RGB images is critic…
arXiv:2607.05389v1 Announce Type: new Abstract: Camera intrinsics are vital for recovering 3D structure from 2D video. However, most 3D algorithms assume fixed intrinsics throughout a video, an assumption that often fails for real-world in-the-wild videos. Consequently, estimatin…
Camera intrinsics are vital for recovering 3D structure from 2D video. However, most 3D algorithms assume fixed intrinsics throughout a video, an assumption that often fails for real-world in-the-wild videos. Consequently, estimating per-frame intrinsics from RGB images is critic…