PulseAugur
中
实时 11:31:55
English(EN) Adaptive Test-Time Memory Allocation for Long-Tailed Human-Object Interaction Detection

新的自适应缓存方法提升了罕见人机交互检测性能

作者 PulseAugur 编辑部 · [1 个来源] ·

研究人员开发了一个自适应多样性缓存（ADC）框架，以改进人机交互（HOI）检测，特别是针对罕见交互。这种无需训练的方法在测试时自适应地分配内存容量，通过频率依赖分配和置信度-多样性联合选择过程来优先处理代表性不足的交互。在HICO-DET和V-COCO数据集上的实验表明，罕见交互的平均精度（mAP）得到了显著提升，增幅在4.02至8.57个百分点之间。 AI

影响 提高了检测罕见人机交互的准确性，可能增强依赖于详细场景理解的AI系统。

排序理由 这是一篇详细介绍计算机视觉新方法的学术论文。[lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.AI 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。 我们如何撰写摘要 →

新的自适应缓存方法提升了罕见人机交互检测性能

本文如何被排名

Signal score
11 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
这是一篇详细介绍计算机视觉新方法的学术论文。[lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
paper, other
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
Same-day
Cluster formed today. Ranking reflects the current source set at time of score.

完整方法见我们的编辑标准。

报道来源 [1]

  1. arXiv cs.AI TIER_1 English(EN) · Yuqiu Jiang, Xiaozhen Qiao, Yifan Chen, Ye Zheng, Zhe Sun, Xuelong Li ·

    面向长尾人机交互检测的自适应测试时内存分配

    arXiv:2511.18811v3 Announce Type: replace-cross Abstract: Long-tailed human-object interaction (HOI) detection remains challenging because rare interactions are sparsely represented in training data. Although test observations provide additional visual evidence, existing cache-ba…

相关实体

相关话题