AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

LocalAgent: Collaborative Agentic Verification for Fine-Grained Instance-Level Consistency

jiachen Guo, Xinshan Zhu, Boyun Wang, Yanyan Liang, Jun Wan, Feng Lin, Ajian Liu

NeurIPS 2026 · Vision-language and multimodal › Multimodal LLMs · approach: Instruction-tuned multimodal assistants · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes LocalAgent, a collaborative agentic framework for fine-grained instance-level consistency verification that checks whether two images show the same physical object under varying viewpoint, illumination, and background.

Problem

MLLMs struggle with fine-grained instance-level consistency verification, deciding whether two images depict the exact same physical object under viewpoint, illumination, and background changes; the task needs both global instance understanding and precise local examination.

Approach

A collaborative agentic verification design that combines global understanding of the target instance with precise local examination of details to judge instance identity; details beyond the abstract are not stated in the provided text.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add LEV tools with auxiliary-capped RFT on top of CoT-RFTQwen3-VL-4B · six-source mean accuracymean accuracy (%): 75.8 -> 80.7Improved
add LEV (LAC+LGC) instead of a crop tool (ablation)Qwen3-VL-4B · six-source meanmean accuracy (%): 75.64 -> 80.70Improved
combine LAC and LGC instead of using a single verifier (ablation)LocalAgent · six-source meanmean accuracy (%): 78.82 (LAC) / 79.94 (LGC) -> 80.70Improved
add weak behavioral prior labels to the tool reward (ablation)LocalAgent · six-source meanmean accuracy (%): 79.39 -> 80.70Improved
use auxiliary cap instead of hard gating (ablation)LocalAgent · six-source meanmean accuracy (%): 80.23 -> 80.70Improved
loosen auxiliary bounds to (0.10, 0.20) (ablation)LocalAgent · six-source meanmean accuracy (%): 80.7 -> 77.1Hurt
remove auxiliary reward on incorrect answers (bounds 0,0) (ablation)LocalAgent · six-source meanmean accuracy (%): 80.7 -> 79.5Hurt
learn selective tool invocation instead of always calling the verifier (ablation)LocalAgent · inference latencymean latency (ms/sample): 610.41 -> 336.86Improved
add LocalAgent selective verification on TWIN-EvalQwen3-VL-4B · TWIN-Eval totalaccuracy (%): 82.9 -> 74.5Hurt
add LocalAgent on training-emphasis-aligned hard TWIN-Eval casesQwen3-VL-4B · TWIN-Eval hard cases subsetaccuracy (%): 72.1 -> 80.9Improved
add LocalAgent on other TWIN-Eval casesQwen3-VL-4B · TWIN-Eval other cases subsetaccuracy (%): 88.8 -> 69.2Hurt
crop to MLLM-localized regions before DINO or XFeat matching (ablation)DINO and XFeat matching · six-source meanmean accuracy (%): 74.23 -> 75.06 (DINO), 69.23 -> 71.11 (XFeat)Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy