Research map / Vision-language and multimodal / Human and embodied modeling
ScaleCap: Scalable Image Captioning via Dual-Modality Debiasing
Long Xing, Qidong Huang, Xiaoyi Dong, Pan Zhang, Yuhang Zang, Yuhang Cao, Jinsong Li, Shuangrui Ding, Weiming Zhang, Nenghai Yu, Jiaqi Wang, Feng Wu and 1 more
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 |
|---|---|---|---|
| further pretrain on ScaleCap-450k captions instead of DenseFusion-450k | LLaVA-NeXT style LVLM · 11-benchmark average, Qwen2.5-7B | average score: 63.0 -> 64.7 | Improved |
| further pretrain with ScaleCap captions on the same DenseFusion images (ablation) | Qwen2.5-3B LVLM · 11-benchmark average | average score: 58.9 -> 60.2 | Improved |
| add both object and position instructions (vs object only) (ablation) | ScaleCap · Prism subset avg | average: 58.8 -> 61.5 | Improved |
| add object instructions (vs position only) (ablation) | ScaleCap · Prism subset avg | average: 57.4 -> 61.5 | Improved |
| scale summarization LLM from Qwen2-7B to Qwen2-72B (ablation) | ScaleCap · MMVet | score: 43.6 -> 58.8 | Improved |
| scale LVLM from 7B to 72B (ablation) | ScaleCap · Prism subset avg | average: 58.5 -> 58.7 | No clear effect |
| scale LVLM from 2B to 7B (ablation) | ScaleCap · Prism subset avg | average: 50.2 -> 58.5 | Improved |
| increase number of heuristic instructions N (ablation) | ScaleCap · MMVet (Prism) | accuracy: 51.8 -> 58.8 | Improved |
| add golden-sentence contrastive filtering (ablation) | LLaVA-v1.5 7B · CHAIR | CHAIR_S: 48.8 -> 33.6 | Improved |
| add golden-sentence contrastive filtering (ablation) | Qwen2-VL 7B · CHAIR | CHAIR_S: 44.2 -> 25.8 | Improved |
| use ScaleCap captions with Qwen2-VL-7B instead of Qwen2-VL-72B generic captions | Prism pipeline · Prism 6-benchmark average | average accuracy: 56.0 -> 58.2 | Improved |
| ScaleCap pretraining vs DenseFusion on ChartQA/MMMU-like individual benchmarks | Qwen2.5-7B LVLM · MMStar and MMMU | score: MMStar 49.2 -> 48.8; MMMU 44.3 -> 46.1 | Mixed |
Closest papers by meaning
- CapsFusion: Rethinking Image-Text Data at Scale (CVPR 2024 · similarity 0.89)
- CaptionQA: Is Your Caption as Useful as the Image Itself? (CVPR 2026 · similarity 0.89)
- CAPability: A Comprehensive Visual Caption Benchmark for Evaluating Both Correctness and Thoroughness (NeurIPS 2025 · similarity 0.89)
- Patch Matters: Training-free Fine-grained Image Caption Enhancement via Local Perception (CVPR 2025 · similarity 0.89)
- FINECAPTION: Compositional Image Captioning Focusing on Wherever You Want at Any Granularity (CVPR 2025 · similarity 0.88)
- MM-SeR: Multimodal Self-Refinement for Lightweight Image Captioning (CVPR 2026 · 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.
