Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

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

About

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).

Lihao Zheng, Zhenwei Shao, Yu Zhou, Yan Yang, Xintian Shen, Jiawei Chen, Hao Ma, Tao Wei• 2026

Related benchmarks

TaskDatasetResultRank
Multi-discipline Multimodal UnderstandingMMMU--
363
Multi-image ReasoningMuirBench
Accuracy66.42
89
Visual Perception and ReasoningBLINK
Accuracy70.02
64
Visual Mathematical ReasoningMathVista
Score78.2
46
Multi-image GroundingMIG-Bench 1.0 (full)
Static Score55.3
19
Fine-grained Multi-Image UnderstandingVLM2-Bench
Matching Score48.85
18
Vision-Language EvaluationMMStar
Average Score72.6
16
Hallucination EvaluationHallusion
Score72.56
12
Showing 8 of 8 rows

Other info

Follow for update