Mimic Human Cognition, Master Multi-Image Reasoning: A Meta-Action Framework for Enhanced Visual Understanding

• URL: http://arxiv.org/abs/2601.07298v1
• Date: Mon, 12 Jan 2026 08:15:36 GMT
• Title: Mimic Human Cognition, Master Multi-Image Reasoning: A Meta-Action Framework for Enhanced Visual Understanding
• Authors: Jianghao Yin, Qingbin Li, Kun Sun, Cheng Ding, Jie Wang, Qin Chen, Jie Zhou, Nan Wang, Changqing Li, Pei Wu, Jian Xu, Zheming Yang, Liang He,
• Abstract summary: We propose a novel approach that decomposes multi-image reasoning into five structured meta-actions: Global, Focus, Hint, Think, and Answer.<n>Our model surpasses GPT-4o on the MUIR and MVMath benchmarks and notably outperforms specialized video reasoning models on video understanding benchmarks.
• Score: 25.81793866721208
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: While Multimodal Large Language Models (MLLMs) excel at single-image understanding, they exhibit significantly degraded performance in multi-image reasoning scenarios. Multi-image reasoning presents fundamental challenges including complex inter-relationships between images and scattered critical information across image sets. Inspired by human cognitive processes, we propose the Cognition-Inspired Meta-Action Framework (CINEMA), a novel approach that decomposes multi-image reasoning into five structured meta-actions: Global, Focus, Hint, Think, and Answer which explicitly modeling the sequential cognitive steps humans naturally employ. For cold-start training, we introduce a Retrieval-Based Tree Sampling strategy that generates high-quality meta-action trajectories to bootstrap the model with reasoning patterns. During reinforcement learning, we adopt a two-stage paradigm: an exploration phase with Diversity-Preserving Strategy to avoid entropy collapse, followed by an annealed exploitation phase with DAPO to gradually strengthen exploitation. To train our model, we construct a dataset of 57k cold-start and 58k reinforcement learning instances spanning multi-image, multi-frame, and single-image tasks. We conduct extensive evaluations on multi-image reasoning benchmarks, video understanding benchmarks, and single-image benchmarks, achieving competitive state-of-the-art performance on several key benchmarks. Our model surpasses GPT-4o on the MUIR and MVMath benchmarks and notably outperforms specialized video reasoning models on video understanding benchmarks, demonstrating the effectiveness and generalizability of our human cognition-inspired reasoning framework.

Related papers

• MICON-Bench: Benchmarking and Enhancing Multi-Image Context Image Generation in Unified Multimodal Models [89.89575486159795]
  We introduce textbfMICON-Bench, a benchmark for multi-image context generation.<n>We propose an MLLM-driven Evaluation-by-Checkpoint framework for automatic verification of semantic and visual consistency.<n>We also present textbfDynamic Attention Rebalancing (DAR), a training-free, plug-and-play mechanism that dynamically adjusts attention during inference to enhance coherence and reduce hallucinations.
  arXiv  Detail & Related papers  (2026-02-23T04:32:52Z)
• More Images, More Problems? A Controlled Analysis of VLM Failure Modes [80.64323947730905]
  Large Vision Language Models (LVLMs) have demonstrated remarkable capabilities, yet their proficiency in understanding and reasoning over multiple images remains largely unexplored.<n>We introduce MIMIC, a new benchmark designed to rigorously evaluate the multi-image capabilities of LVLMs.
  arXiv  Detail & Related papers  (2026-01-12T18:45:13Z)
• Thinking with Images for Multimodal Reasoning: Foundations, Methods, and Future Frontiers [90.4459196223986]
  A similar evolution is now unfolding in AI, marking a paradigm shift from models that merely think about images to those that can truly think with images.<n>This emerging paradigm is characterized by models leveraging visual information as intermediate steps in their thought process, transforming vision from a passive input into a dynamic, manipulable cognitive workspace.
  arXiv  Detail & Related papers  (2025-06-30T14:48:35Z)
• PeRL: Permutation-Enhanced Reinforcement Learning for Interleaved Vision-Language Reasoning [50.21619363035618]
  We propose a general reinforcement learning approach PeRL tailored for interleaved multimodal tasks.<n>We introduce permutation of image sequences to simulate varied positional relationships to explore more spatial and positional diversity.<n>Our experiments confirm that PeRL trained model consistently surpasses R1-related and interleaved VLM baselines by a large margin.
  arXiv  Detail & Related papers  (2025-06-17T18:25:56Z)
• Weaving Context Across Images: Improving Vision-Language Models through Focus-Centric Visual Chains [31.828341309787042]
  Vision-language models (VLMs) achieve remarkable success in single-image tasks.<n>Real-world scenarios often involve intricate multi-image inputs, leading to a notable performance decline.<n>We propose Focus-Centric Visual Chain, a novel paradigm that enhances VLMs'perception, comprehension, and reasoning abilities in multi-image scenarios.
  arXiv  Detail & Related papers  (2025-04-28T19:02:18Z)
• CMMCoT: Enhancing Complex Multi-Image Comprehension via Multi-Modal Chain-of-Thought and Memory Augmentation [12.008690947774015]
  We propose a multi-step reasoning framework that mimics human-like "slow thinking" for multi-image understanding.<n>The construction of interleaved multimodal multi-step reasoning chains, which utilize critical visual region tokens, as supervisory signals.<n>The introduction of a test-time memory augmentation module that expands the model reasoning capacity during inference.
  arXiv  Detail & Related papers  (2025-03-07T09:13:17Z)
• MMIU: Multimodal Multi-image Understanding for Evaluating Large Vision-Language Models [76.1999277491816]
  Multimodal Multi-image Understanding (MMIU) is a comprehensive evaluation suite designed to assess Large Vision-Language Models (LVLMs) MMIU encompasses 7 types of multi-image relationships, 52 tasks, 77K images, and 11K meticulously curated multiple-choice questions. Our evaluation of 24 popular LVLMs, including both open-source and proprietary models, reveals significant challenges in multi-image comprehension.
  arXiv  Detail & Related papers  (2024-08-05T17:56:41Z)
• Benchmarking Multi-Image Understanding in Vision and Language Models: Perception, Knowledge, Reasoning, and Multi-Hop Reasoning [15.296263261737026]
  We introduce a Multi-Image MIRB Benchmark to evaluate visual language models' ability to compare, analyze, and reason across multiple images. Our benchmark encompasses four categories: perception, visual world knowledge, reasoning, and multi-hop reasoning. We demonstrate that while open-source VLMs were shown to approach the GPT-4V in single-image tasks, a significant gap remains in multi-image reasoning tasks.
  arXiv  Detail & Related papers  (2024-06-18T16:02:18Z)
• MM1: Methods, Analysis & Insights from Multimodal LLM Pre-training [103.72844619581811]
  We build performant Multimodal Large Language Models (MLLMs) In particular, we study the importance of various architecture components and data choices. We demonstrate that for large-scale multimodal pre-training using a careful mix of image-caption, interleaved image-text, and text-only data.
  arXiv  Detail & Related papers  (2024-03-14T17:51:32Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.