研究人员推出了一种新颖的自监督学习方法 Cross4D-JEPA，用于理解动态 4D 点云。该方法将来自 DINOv2 和 V-JEPA 2 等 2D 图像或视频基础模型的知识蒸馏到一个 4D 点编码器中。Cross4D-JEPA 利用密集的跨模态对应关系将 3D 点映射到教师块特征，训练学生编码器以匹配这些特征，而无需掩码、负样本或解码器。与单模态和全局跨模态基线相比，该方法在 MSR-Action3D 和 NTU RGB+D 60 等基准测试中表现出优越的性能，突显了其细粒度对应方法的有效性。 AI
影响 增强了 4D 点云分析的自监督学习，可能改进机器人和具身感知。
排序理由 介绍新方法及其评估的学术论文。[lever_c_demoted from research: ic=1 ai=1.0]
AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →
arXiv:2607.00514v1 Announce Type: cross Abstract: Automatic understanding of dynamic 4D point clouds, the 3D-point sequences captured over time by depth sensors and LiDAR, is central to robotics and embodied perception. Yet annotating them densely is expensive, making self-superv…
Automatic understanding of dynamic 4D point clouds, the 3D-point sequences captured over time by depth sensors and LiDAR, is central to robotics and embodied perception. Yet annotating them densely is expensive, making self-supervised pretraining the natural route to transferable…