mlsanityData to 14 Oct 2026

CGC: Compositional Grounded Contrast for Fine-Grained Multi-Image Understanding

arXiv 2604.22498 ↗First posted 24 Apr 2026cs.CV cs.AIcited by 0

On the map: Multimodal LLMs / vision-language models → VLM benchmarks and visual reasoning

Abstract (arXiv)

Although Multimodal Large Language Models (MLLMs) have advanced rapidly, they still face notable challenges in fine-grained multi-image understanding, often exhibiting spatial hallucination, attention leakage, and failures in object constancy. In addition, existing approaches typically rely on expensive human annotations or large-scale chain-of-thought (CoT) data generation. We propose Compositional Grounded Contrast (abbr. CGC), a low-cost full framework for boosting fine-grained multi-image understanding of MLLMs. Built on existing single-image grounding annotations, CGC constructs compositional multi-image training instances through Inter-Image Contrast and Intra-Image Contrast, which introduce semantically decoupled distractor contexts for cross-image discrimination and correlated cross-view samples for object constancy, respectively. CGC further introduces a Rule-Based Spatial Reward within the GRPO framework to improve source-image attribution, spatial alignment, and structured output validity under a Think-before-Grounding paradigm. Experiments show that CGC achieves state-of-the-art results on fine-grained multi-image benchmarks, including MIG-Bench and VLM2-Bench. The learned multi-image understanding capability also transfers to broader multimodal understanding and reasoning tasks, yielding consistent gains over the Qwen3-VL-8B base model on MathVista (+2.90), MuirBench (+2.88), MMStar (+1.93), MMMU (+1.77), and BLINK (+1.69).

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in VLM benchmarks and visual reasoning, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 29 of its 51 references that are in the map. First posted on arXiv 24 Apr 2026.

Builds on

29 of 51 references.

Closest ancestors · 3

Same direction, newest first

  1. 2025Migician: Revealing the Magic of Free-Form Multi-Image Grounding in Multimodal Large Language Models10 Jan
  2. 2023Ferret: Refer and Ground Anything Anywhere at Any Granularity11 Oct
  3. 2023LISA: Reasoning Segmentation via Large Language Model1 Aug

Other references · 26

  1. 2025Qwen2.5-VL Technical Reportcited by 5,587
  2. 2024LLaVA-OneVision: Easy Visual Task Transfercited by 2,719
  3. 2024Expanding Performance Boundaries of Open-Source Multimodal Models with Model, Data, and Test-Time Scalingcited by 1,599
  4. 2024MiniCPM-V: A GPT-4V Level MLLM on Your Phonecited by 1,079
  5. 2016Modeling Context in Referring Expressionscited by 1,063
  6. 2023Shikra: Unleashing Multimodal LLM's Referential Dialogue Magiccited by 906
  7. 2025V-JEPA 2: Self-Supervised Video Models Enable Understanding, Prediction and Planningcited by 809
  8. 2024BLINK: Multimodal Large Language Models Can See but Not Perceivecited by 550
  9. 2023mPLUG-Owl2: Revolutionizing Multi-modal Large Language Model with Modality Collaborationcited by 547
  10. 2024Long Context Transfer from Language to Visioncited by 515
  11. 2025Visual-RFT: Visual Reinforcement Fine-Tuningcited by 513
  12. 2025R1-VL: Learning to Reason with Multimodal Large Language Models via Step-wise Group Relative Policy Optimizationcited by 301
14 more
  1. 2024MANTIS: Interleaved Multi-Image Instruction Tuningcited by 216
  2. 2025SFT or RL? An Early Investigation into Training R1-Like Reasoning Large Vision-Language Modelscited by 212
  3. 2025SoTA with Less: MCTS-Guided Sample Selection for Data-Efficient Visual Reasoning Self-Improvementcited by 145
  4. 2025Perception-R1: Pioneering Perception Policy with Reinforcement Learningcited by 120
  5. 2021Learning to Predict Visual Attributes in the Wildcited by 96
  6. 2024MIBench: Evaluating Multimodal Large Language Models over Multiple Imagescited by 39
  7. 2023Small Object Detection via Coarse-to-fine Proposal Generation and Imitation Learningcited by 27
  8. 2025Are They the Same? Exploring Visual Correspondence Shortcomings of Multimodal LLMscited by 16
  9. 2025DIP-R1: Deep Inspection and Perception with RL Looking Through and Understanding Complex Scenescited by 11
  10. 2025MindGPT-4ov: An Enhanced MLLM via a Multi-Stage Post-Training Paradigmcited by 9
  11. 2026Evolving from Tool User to Creator via Training-Free Experience Reuse in Multimodal Reasoningcited by 7
  12. 2025MIRG-RL: Multi-Image Reasoning and Grounding with Reinforcement Learningcited by 3
  13. 2026DIVA-GRPO: Enhancing Multimodal Reasoning through Difficulty-Adaptive Variant Advantagecited by 3
  14. 2026Evaluating the Search Agent in a Parallel Worldcited by 2

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2024CoF: Coarse to Fine-Grained Image Understanding for Multi-modal Large Language Models
  2. 2026GeM-VG: Towards Generalized Multi-image Visual Grounding with Multimodal Large Language Models
  3. 2024Advancing Fine-Grained Visual Understanding with Multi-Scale Alignment in Multi-Modal Models
  4. 2023Pink: Unveiling the Power of Referential Comprehension for Multi-modal LLMs
  5. 2026Think with Structured Grounding: Perceptual Reinforcement Learning for Chart and Visual-Tabular Understanding
  6. 2025DiG: Differential Grounding for Enhancing Fine-Grained Perception in Multimodal Large Language Model
  7. 2025Grounded Chain-of-Thought for Multimodal Large Language Models
  8. 2025Hierarchical Contextual Grounding LVLM: Enhancing Fine-Grained Visual-Language Understanding with Robust Grounding