发表机构
ELLIS Institute Finland; University of Oulu（芬兰ELLIS研究所; 奥卢大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
该研究提出Human-JEPA模型，以锚定预测方式在视频上训练，解决现有以人类为中心视觉模型无法实现运动与预测的问题，在姿态和人物重识别任务上参数更少且性能更优，可同时满足理解人类的两大需求。
AI 中文摘要
理解人类的机器应能感知当下并预测未来。现有的以人类为中心的视觉模型均在人类图像上进行预训练，在静态密集感知方面达到了顶尖水平，但无法实现运动与预测功能。本文提出Human-JEPA，这是一种以人类为中心的视觉模型，通过锚定预测的方式在视频上进行训练：将密集目标固定在初始化的冻结副本上，防止密集感知的静默崩溃；同时将块掩码替换为纯过去到未来的划分，避免了5点动作分类和17点重识别的崩溃。在冻结探针设置下，Human-JEPA在姿态和人物重识别任务上，以仅2.7倍更少的参数数量领先于基于像素锚定的专用模型，仅在高分辨率密集解析方面略有让步；其发布的预测器头部是首个不会降低预测性能的模型。因此，一个经安全适配的单一模型即可满足理解人类的两大需求。
英文摘要
Machines that understand humans should perceive the present and anticipate the future. Existing human-centric vision model are pretrained on human images, set the state of the art in static dense perception, so motion and anticipation are out of reach. Here we present Human-JEPA, a human-centric vision model trained on video by anchored forecasting: dense targets are pinned to a frozen copy of the initialization, preventing a silent collapse of dense perception, and block masks are replaced by a pure past-to-future split, avoiding a five-point action tax and a seventeen-point re-identification collapse. Under frozen probes, Human-JEPA leads the pixel-anchored specialists on pose and person re-identification at 2.7 times fewer parameters, conceding high-resolution dense parsing, and its released predictor head is the first that does not degrade anticipation. A single safely adapted model thus serves both halves of understanding humans.
