Skip to content
Hot eventLive

VICO:视觉环境与VLM协同进化框架

1 reports1 sources3 hr ago updated

Get the story

AI overview

2026年10月9日,arXiv Computer Vision 发布论文,提出 VICO 框架,旨在让视觉环境与视觉语言模型(VLM)智能体协同进化。该框架引入 Environment-as-Rewriter(EnvRewriter),通过编辑场景图、图表表格或保护区域掩码并重新渲染,生成与模型当前能力难度相匹配的样本。目前公开信息仅涉及该框架的核心思路与机制,尚未披露实验结果、数据集或开源情况。

Generated from reports · updated 3 hr ago

Timeline

Follow the coverage from different angles.

Oct 9, 2026
  1. arXiv · Computer Vision
    VICO:视觉环境与视觉语言模型协同进化的推理框架

    VICO 提出让视觉环境与 VLM 智能体协同进化,通过 Environment-as-Rewriter(EnvRewriter)编辑场景图、图表表格或保护区域掩码并重新渲染,生成难度与模型当前能力匹配的样本。

Heat trend

Current heat 9·Comparable peak 10(Oct 9)·Comparable change over 24 hours –

02.557.510Oct9Oct9Oct9Oct9

The trend compares only the same participants observed continuously; its range may be smaller than the current heat count. Move or click on the chart to inspect hourly heat; use the left and right arrow keys to switch.