看得更远，想得更深：以低层视觉线索与反思提升视觉语言模型的推理能力
See Further, Think Deeper: Advancing VLM's Reasoning Ability with Low-level Visual Cues and Reflection
把低层视觉工具（如显著性/掩码提取）接入 VLM 的思维链，并用基于掩码的视觉反馈让模型在推理中回看图像、修正答案；对做视觉推理、细粒度视觉定位或多模态 agent 的研究者，这是一套可复用的「工具调用+视觉反思」RL 训练范式。
原文依据 a mask-based visual feedback mechanism is elaborated to incorporate visual reflection into the thinking process
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论近期视觉语言模型（VLM）的进展得益于强化学习（RL）带来的推理能力增强。然而，现有方法仍面临关键局限，包括缺乏低层视觉信息以及缺乏有效的视觉反馈。为解决这些问题，本文提出一个统一的多模态交错推理框架 ForeSight，使 VLM 能够借助低层视觉线索「看得更远」，并借助有效的视觉反馈「想得更深」。首先，该框架引入一组低层视觉工具，将关键视觉信息整合进推理链，缓解对细粒度视觉特征的忽视。其次，本文设计了一种基于掩码的视觉反馈机制，将视觉反思纳入思考过程，使模型能够动态地重新审视并更新其答案。在强化学习的驱动下，ForeSight 学会自主决定工具调用与答案验证，并以最终答案准确率作为奖励信号。为评估所提框架的性能，本文基于 SalBench 数据集构建了一个新数据集 Character and Grounding SalBench（CG-SalBench）。实验结果表明，ForeSight-7B 模型显著优于同等参数规模的其他模型，在某些指标上甚至超越了当前最先进的闭源模型。
英文原文摘要 1370 字符
Recent advances in Vision-Language Models (VLMs) have benefited from Reinforcement Learning (RL) for enhanced reasoning. However, existing methods still face critical limitations, including the lack of low-level visual information and effective visual feedback. To address these problems, this paper proposes a unified multimodal interleaved reasoning framework ForeSight, which enables VLMs to See Further with low-level visual cues and Think Deeper with effective visual feedback. First, it introduces a set of low-level visual tools to integrate essential visual information into the reasoning chain, mitigating the neglect of fine-grained visual features. Second, a mask-based visual feedback mechanism is elaborated to incorporate visual reflection into the thinking process, enabling the model to dynamically re-examine and update its answers. Driven by RL, ForeSight learns to autonomously decide on tool invocation and answer verification, with the final answer accuracy as the reward signal. To evaluate the performance of the proposed framework, we construct a new dataset, Character and Grounding SalBench (CG-SalBench), based on the SalBench dataset. Experimental results demonstrate that the ForeSight-7B model significantly outperforms other models with the same parameter scale, and even surpasses the current SOTA closed-source models on certain metrics.