研究人员开发了用于3D人体姿态估计的新方法，其中一项研究侧重于2D预训练在提高计算效率和跨数据集泛化能力方面的优势。该方法在MPII和Human3.6M等基准测试上的表现持续优于仅使用3D数据训练的方法，并达到了特定的性能指标。另一篇论文介绍了一个用于多视角姿态估计的非标定框架，该框架利用深度神经网络、代数先验和时间动态，无需精确的相机标定即可工作，为此类方法设定了新的最先进水平。 AI
影响 非标定多视角姿态估计和高效2D预训练的进步可以为更强大、更易于访问的3D动作捕捉应用提供支持。
排序理由 该集群包含两篇arXiv论文，详细介绍了计算机视觉领域的新研究，特别是3D人体姿态估计。
AI 生成摘要 · Google Gemini · 来自 4 个来源。 我们如何撰写摘要 →
Recovering 3D human pose from multi-view imagery typically relies on precise camera calibration, which is often unavailable in real-world scenarios, thereby severely limiting the applicability of existing methods. To overcome this challenge, we propose an unconstrained framework …
arXiv:2604.22830v1 Announce Type: new Abstract: Pre-training is a general method that is used in a range of deep learning tasks. By first training a model on one task, and then further training on the downstream task used for final evaluation, the model is forced to learn a more …
arXiv:2604.24312v1 Announce Type: new Abstract: Recovering 3D human pose from multi-view imagery typically relies on precise camera calibration, which is often unavailable in real-world scenarios, thereby severely limiting the applicability of existing methods. To overcome this c…
Recovering 3D human pose from multi-view imagery typically relies on precise camera calibration, which is often unavailable in real-world scenarios, thereby severely limiting the applicability of existing methods. To overcome this challenge, we propose an unconstrained framework …