RSAgent 基于多轮工具调用,学习为文本引导分割任务进行推理与执行 RSAgent Learning to Reason and Act for Text-Guided Segmentation via Multi-Turn Tool Invocations.pdfVIP

  • 1
  • 0
  • 约8.7万字
  • 约 14页
  • 2026-08-27 发布于广东
  • 举报

RSAgent 基于多轮工具调用,学习为文本引导分割任务进行推理与执行 RSAgent Learning to Reason and Act for Text-Guided Segmentation via Multi-Turn Tool Invocations.pdf

RSAgent:LearningtoReasonandActforText-GuidedSegmentation

viaMulti-TurnToolInvocations

XingqiHe†1YujieZhang†34ShuyongGao*1WenjieLi45LingyiHong1MingxiChen1KaixunJiang2

JiyuanFu1WenqiangZhang*12

Abstract

5

2

0

2Text-guidedobjectsegmentationrequiresboth

cross-modalreasoningandpixelgroundingabil-

c

eities.Mostr

您可能关注的文档

文档评论(0)

1亿VIP精品文档

相关文档