研究人员开发了一个名为MetricScenes的新数据集，以解决单目几何估计中的“尺度坍塌”问题，即远处物体表示不准确。该数据集由互联网照片和立体图像编译而成，提供了在真实环境中具有度量基础的场景。在MetricScenes上微调MoGe-2模型可显著提高其在无约束环境中估计绝对尺度的准确性。 AI
影响 提高了从单目图像理解3D场景的能力，可能有助于机器人和增强现实等应用。
排序理由 该集群包含一篇学术论文，详细介绍了用于特定计算机视觉任务的新数据集和模型微调。
在 Hugging Face Daily Papers 阅读 →
AI 生成摘要 · Google Gemini · 来自 3 个来源。 我们如何撰写摘要 →
Metric scale monocular geometry estimation has seen significant progress through large-scale data aggregation, yet current foundation models suffer from a persistent ''scale-collapse'' phenomenon: distant landmarks and vast landscapes are metrically underestimated. We hypothesize…
arXiv:2606.02379v1 Announce Type: new Abstract: Metric scale monocular geometry estimation has seen significant progress through large-scale data aggregation, yet current foundation models suffer from a persistent ''scale-collapse'' phenomenon: distant landmarks and vast landscap…
Metric scale monocular geometry estimation has seen significant progress through large-scale data aggregation, yet current foundation models suffer from a persistent ''scale-collapse'' phenomenon: distant landmarks and vast landscapes are metrically underestimated. We hypothesize…