Research map / Vision-language and multimodal / Human and embodied modeling
UniTAB: Unifying Text and Box Outputs for Grounded Vision-Language Modeling
Zhengyuan Yang, Zhe Gan, Jianfeng Wang, Xiaowei Hu, Faisal Ahmed, Zicheng Liu, Yumao Lu, Lijuan Wang
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 pre-training (ablation) | UniTAB (task-specific finetuning) · Refcocog val-u | Acc@0.5: 63.56 -> 79.96 | Improved |
| add pre-training (ablation) | UniTAB (multi-task finetuning) · Flickr30k grounded captioning | CIDEr: 61.1 -> 63.4 | Improved |
| replace task-specific finetuning by multi-task finetuning (Shared) after pre-training (ablation) | UniTAB · Refcocog val-u | Acc@0.5: 79.96 -> 84.46 | Improved |
| replace task-specific finetuning by multi-task finetuning (Shared) after pre-training (ablation) | UniTAB · Flickr30k grounded captioning | CIDEr: 65.6 -> 63.4 | Hurt |
| add task-specific finetuning after multi-task finetuning (Pre-finetuning) (ablation) | UniTAB Shared · Flickr30k grounded captioning | CIDEr: 63.4 -> 69.7 | Improved |
| add 4M image-text pairs without boxes to pre-training (ablation) | UniTAB Separate · Flickr30k grounded captioning | CIDEr: +8.6 CIDEr (65.6 -> 74.2) | Improved |
| add 4M image-text pairs without boxes to pre-training (ablation) | UniTAB Separate · COCO captioning | CIDEr: +3.8 CIDEr (119.3 -> 123.1) | Improved |
| add 4M image-text pairs without boxes to pre-training (ablation) | UniTAB Separate · VQAv2 Karpathy test | accuracy: +2.5% absolute (66.6 -> 69.1) | Improved |
| add extra detection box annotations to pre-training (ablation) | UniTAB Separate · Refcocog val-u | Acc@0.5: 79.96 -> 83.78 | Improved |
| add extra detection box annotations to pre-training (ablation) | UniTAB Separate · Flickr30k grounded captioning | CIDEr: 65.6 -> 70.0 | Improved |
| add obj tokens around word-box segments (ablation) | UniTAB Separate · Refcocog val-u | Acc@0.5: 79.31 -> 80.23 | Improved |
| add decoder type embedding (ablation) | UniTAB Separate · Refcocog val-u | Acc@0.5: 80.23 -> 80.54 | No clear effect |
Later papers that compared with this method
1 later accepted paper reports UniTAB as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Region-aware Anchoring Mechanism for Efficient Referring Visual Grounding (ICCV 2025 · better on 3 of 3 shared results)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- MDETR — MDETR - Modulated Detection for End-to-End Multi-Modal Understanding (ICCV 2021) · better on 3 of 4 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Your Large Vision-Language Model Only Needs A Few Attention Heads For Visual Grounding (CVPR 2025 · similarity 0.89)
- Connecting the Dots: Training-Free Visual Grounding via Agentic Reasoning (AAAI 2026 · similarity 0.88)
- GroundingME: Exposing the Visual Grounding Gap in MLLMs through Multi-Dimensional Evaluation (CVPR 2026 · similarity 0.88)
- RegionGPT: Towards Region Understanding Vision Language Model (CVPR 2024 · similarity 0.88)
- PostAlign: Multimodal Grounding as a Corrective Lens for MLLMs (ICLR 2026 · similarity 0.87)
- SimVG: A Simple Framework for Visual Grounding with Decoupled Multi-modal Fusion (NeurIPS 2024 · similarity 0.87)
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.
