Research map / Vision-language and multimodal / Multimodal LLMs
DiCoBench: Benchmarking Multi-Image Fine-Grained Perception via Differential and Commonality Visual Cues
Geng Li, Yuxin Peng
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| compare humans against best MLLM on cross-image perception | Gemini-3-Pro · DiCoBench | average accuracy: 98.3 -> 58.1 | Hurt |
| give humans unlimited time instead of tight time limits (ablation) | human annotators · DiCoBench | accuracy | Improved |
| crop input to annotated cue region (ablation) | Qwen3-VL-8B · DiCoBench (1/10 subset) | accuracy | Improved |
| crop then resize back to original dimensions (ablation) | Qwen3-VL-8B · DiCoBench (1/10 subset) | accuracy | Improved |
| use lightweight task-specific tuned model (TreeVGR-7B) vs larger general closed model | TreeVGR-7B · DiCoBench | average accuracy: 29.0 -> 48.8 | Improved |
| scale open model from Qwen2.5-VL-7B to 32B (ablation) | Qwen2.5-VL · DiCoBench | average accuracy: 41.0 -> 31.4 | Hurt |
Closest papers by meaning
- Zooming without Zooming: Region-to-Image Distillation for Fine-Grained Multimodal Perception (ICML 2026 · similarity 0.91)
- OddGridBench: Exposing the Lack of Fine-Grained Visual Discrepancy Sensitivity in Multimodal Large Language Models (CVPR 2026 · similarity 0.90)
- Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models (AAAI 2025 · similarity 0.90)
- JourneyBench: A Challenging One-Stop Vision-Language Understanding Benchmark of Generated Images (NeurIPS 2024 · similarity 0.90)
- DiG: Differential Grounding for Enhancing Fine-Grained Perception in Multimodal Large Language Models (CVPR 2026 · similarity 0.89)
- Benchmarking Large Vision-Language Models on Fine-Grained Image Tasks: A Comprehensive Evaluation (ICLR 2026 · similarity 0.89)
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.
