Research map / Vision-language and multimodal / Multimodal LLMs
Advancing Visual Large Language Model for Multi-granular Versatile Perception
Wentao Xiang, Haoxian Tan, Yujie Zhong, Cong Wei, Dengjie Li, Yujiu Yang
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 O365 data to joint training (ablation) | MVP-LM · RefCOCO val | cIoU: 77.6 -> 81.8 | Improved |
| add Grounding data (GoldG) after O365 (ablation) | MVP-LM · RefCOCO val | cIoU: 81.8 -> 83.6 | Improved |
| add O365 data to joint training (ablation) | MVP-LM · COCO-Panoptic | PQ: 56.4 -> 55.8 | Hurt |
| add Grounding data after O365 (ablation) | MVP-LM · COCO-Panoptic | PQ: 55.8 -> 56.1 | Improved |
| generate image caption before summary token (ablation) | MVP-LM · COCO-Panoptic | PQ: 55.3 -> 55.6 | Improved |
| generate image caption before summary token (ablation) | MVP-LM · COCO-Panoptic | mIoU: 65.7 -> 66.2 | Improved |
| generate image caption before summary token (ablation) | MVP-LM · RefCOCO val | cIoU: 75.6 -> 75.7 | No clear effect |
| generate existing object names before summary token (ablation) | MVP-LM · COCO-Panoptic | PQ: 55.3 -> 54.9 | Hurt |
| use similar visual features directly as final queries (no LLM base query) (ablation) | MVP-LM · RefCOCO val | cIoU: 75.7 -> 75.9 | Improved |
| use similar visual features directly as final queries (no LLM base query) (ablation) | MVP-LM · COCO-Panoptic | PQ: 55.6 -> 55.0 | Hurt |
| increase queries from 30 to 300 (ablation) | MVP-LM · RefCOCO val | cIoU: 76.3 -> 72.4 | Hurt |
| use 100 queries instead of 30 (ablation) | MVP-LM · COCO-Panoptic | PQ: 53.1 -> 55.6 | Improved |
Closest papers by meaning
- VisionLLM v2: An End-to-End Generalist Multimodal Large Language Model for Hundreds of Vision-Language Tasks (NeurIPS 2024 · similarity 0.89)
- Large Vision-Language Models Get Lost in Attention (ICML 2026 · similarity 0.89)
- PerceptionGPT: Effectively Fusing Visual Perception into LLM (CVPR 2024 · similarity 0.89)
- Benchmarking Large Vision-Language Models on Fine-Grained Image Tasks: A Comprehensive Evaluation (ICLR 2026 · similarity 0.89)
- DenseFusion-1M: Merging Vision Experts for Comprehensive Multimodal Perception (NeurIPS 2024 · similarity 0.89)
- Synthesize Diagnose and Optimize: Towards Fine-Grained Vision-Language Understanding (CVPR 2024 · 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.
