English(EN) CIRCLED: A Multi-turn CIR Dataset with Consistent Dialogues across Domains

新的CIRCLED数据集增强了多轮图像检索研究

作者 PulseAugur 编辑部 · [2 个来源] · 2026-05-26 09:11

研究人员推出了一款名为CIRCLED的新型多轮图像检索数据集，旨在克服现有数据集在对话一致性和领域限制（如时尚领域）方面的不足。CIRCLED包含22,608个多轮对话会话，分布在九个子集中，查询会逐步细化以指向目标图像。该数据集旨在为推进多轮图像检索研究提供高质量的基准。 AI

影响为改进多轮图像检索系统提供了一个新基准。

排序理由该集群描述了一个针对特定研究任务的新学术数据集的发布。

AI 生成摘要 · Google Gemini · 来自 2 个来源。我们如何撰写摘要 →

报道来源 [2]

arXiv cs.CV TIER_1 English(EN) · Tomohisa Takeda, Yu-Chieh Lin, Yuji Nozawa, Youyang Ng, Osamu Torii, Yusuke Matsui · 2026-05-27 04:00

CIRCLED：跨领域一致性对话的多轮CIR数据集

arXiv:2605.26734v1 Announce Type: new Abstract: Existing Multi-Turn Composed Image Retrieval (MTCIR) datasets lack dialogue-history consistency and are restricted to the fashion domain. To address these limitations, we construct CIRCLED by extending FashionIQ, CIRR, and CIRCO. In…
arXiv cs.CV TIER_1 English(EN) · Yusuke Matsui · 2026-05-26 09:11

CIRCLED：跨领域一致性对话的多轮CIR数据集

Existing Multi-Turn Composed Image Retrieval (MTCIR) datasets lack dialogue-history consistency and are restricted to the fashion domain. To address these limitations, we construct CIRCLED by extending FashionIQ, CIRR, and CIRCO. In CIRCLED, the query at each turn progressively a…