研究人员推出EgoMAGIC，一个专为医学领域感知算法训练设计的新数据集。该数据集包含3,355个以自我为中心的视频，涵盖50项不同的医疗任务，是作为DARPA的感知任务指导计划的一部分收集的。该数据旨在支持增强现实头显的虚拟助手开发，并已发布一个动作检测挑战。此外，40个YOLO模型在该数据集上进行了训练，使用了195万个标签来识别124个医疗对象。 AI
影响 为开发AI驱动的医疗辅助工具提供了大规模数据集和基线结果。
排序理由 该集群描述了一个用于医学领域计算机视觉任务的新数据集及其相关研究论文的发布。
AI 生成摘要 · Google Gemini · 来自 3 个来源。 我们如何撰写摘要 →
This paper introduces EgoMAGIC (Medical Assistance, Guidance, Instruction, and Correction), an egocentric medical activity dataset collected as part of DARPA's Perceptually-enabled Task Guidance (PTG) program. This dataset comprises 3,355 videos of 50 medical tasks, with at least…
arXiv:2604.22036v1 Announce Type: new Abstract: This paper introduces EgoMAGIC (Medical Assistance, Guidance, Instruction, and Correction), an egocentric medical activity dataset collected as part of DARPA's Perceptually-enabled Task Guidance (PTG) program. This dataset comprises…
This paper introduces EgoMAGIC (Medical Assistance, Guidance, Instruction, and Correction), an egocentric medical activity dataset collected as part of DARPA's Perceptually-enabled Task Guidance (PTG) program. This dataset comprises 3,355 videos of 50 medical tasks, with at least…