Research map / Vision-language and multimodal / Multimodal LLMs
OURO: A Self-Bootstrapped Framework for Enhancing Multimodal Scene Understanding
Tianrun Xu, Guanyu Chen, Ye Li, Yuxin Xi, Zeyu Mu, Ruichen Wang, Tianren Zhang, Haichuan Gao, Feng Chen
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 |
|---|---|---|---|
| add 5 training sub-regions instead of none (ablation) | OURO-7B · Average over 10 VQA benchmarks | average accuracy: 77.3 -> 80.0 | Improved |
| add 1 random sub-region instead of none (ablation) | OURO-7B · Average over 10 VQA benchmarks | average accuracy: 77.3 -> 74.0 | Hurt |
| add 2 sub-regions instead of none (ablation) | OURO-7B · Average over 10 VQA benchmarks | average accuracy: 77.3 -> 78.8 | Improved |
| add 3 sub-regions instead of none (ablation) | OURO-7B · Average over 10 VQA benchmarks | average accuracy: 77.3 -> 78.7 | Improved |
| increase sub-regions from 5 to 7 (ablation) | OURO-7B · Average over 10 VQA benchmarks | average accuracy: 80.0 -> 75.6 | Hurt |
| use enhanced self-bootstrapped annotations instead of original (k=3) (ablation) | OURO-7B · VQAv2 | accuracy: 72.3 -> 79.9 | Improved |
| use enhanced annotations instead of original (k=3) (ablation) | OURO-7B · AI2D | accuracy: 66.0 -> 86.5 | Improved |
| use enhanced annotations instead of original (k=3) (ablation) | OURO-7B · OKVQA | accuracy: 58.4 -> 65.9 | Improved |
| use enhanced annotations instead of original (k=5) (ablation) | OURO-7B · ChartQA | accuracy: 80.2 -> 84.1 | Improved |
| use enhanced annotations instead of original (k=5) (ablation) | OURO-7B · STVQA | accuracy: 72.8 -> 77.0 | Improved |
| use enhanced annotations instead of original (k=5) (ablation) | OURO-7B · AI2D | accuracy: 85.9 -> 80.2 | Hurt |
| use enhanced annotations instead of original (k=5) (ablation) | OURO-7B · OKVQA | accuracy: 66.2 -> 66.2 | No clear effect |
Closest papers by meaning
- MMIU: Multimodal Multi-image Understanding for Evaluating Large Vision-Language Models (ICLR 2025 · similarity 0.91)
- Human-MME: A Holistic Evaluation Benchmark for Human-Centric Multimodal Large Language Models (ICLR 2026 · similarity 0.88)
- Dynamic Multimodal Evaluation with Flexible Complexity by Vision-Language Bootstrapping (ICLR 2025 · similarity 0.88)
- Seeing What Matters: A Training-Free Self-Guided Framework for Multimodal Detail Perception and Reasoning (CVPR 2026 · similarity 0.88)
- UVU: Improving Multimodal Understanding via Vision-Language Unified Autoregressive Paradigm (CVPR 2026 · similarity 0.88)
- MMIE: Massive Multimodal Interleaved Comprehension Benchmark for Large Vision-Language Models (ICLR 2025 · similarity 0.88)
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.
