Research map / Vision-language and multimodal / Human and embodied modeling
TransVG: End-to-End Visual Grounding With Transformers
Jiajun Deng, Zhengyuan Yang, Tianlang Chen, Wengang Zhou, Houqiang Li
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 |
|---|---|---|---|
| use a learnable embedding for the [REG] token instead of average-pooled visual tokens (ablation) | TransVG (ResNet-50) · RefCOCO val | top-1 accuracy (%): 79.12 -> 80.32 | Improved |
| use a learnable embedding for the [REG] token instead of max-pooled visual tokens (ablation) | TransVG (ResNet-50) · RefCOCO val | top-1 accuracy (%): 78.37 -> 80.32 | Improved |
| use a learnable embedding for the [REG] token instead of average-pooled linguistic tokens (ablation) | TransVG (ResNet-50) · RefCOCO val | top-1 accuracy (%): 78.51 -> 80.32 | Improved |
| use a learnable embedding for the [REG] token instead of max-pooled linguistic tokens (ablation) | TransVG (ResNet-50) · RefCOCO val | top-1 accuracy (%): 78.74 -> 80.32 | Improved |
| use a learnable embedding for the [REG] token instead of the linguistic [CLS] token (ablation) | TransVG (ResNet-50) · RefCOCO val | top-1 accuracy (%): 77.90 -> 80.32 | Improved |
| add visual transformer (ablation) | TransVG (ResNet-50) · ReferItGame test | top-1 accuracy (%): 64.24 -> 68.48 | Improved |
| add linguistic transformer (ablation) | TransVG (ResNet-50) · ReferItGame test | top-1 accuracy (%): 64.24 -> 66.78 | Improved |
| add both visual and linguistic transformers (ablation) | TransVG (ResNet-50) · ReferItGame test | top-1 accuracy (%): 64.24 -> 69.76 | Improved |
| replace ResNet-50 with ResNet-101 backbone | TransVG · ReferItGame test | top-1 accuracy (%): 69.76 -> 70.73 | Improved |
| replace multi-round fusion of ReSC-Large with transformer encoder fusion | TransVG (ResNet-101) · ReferItGame test | top-1 accuracy (%): 64.60 -> 70.73 | Improved |
Later papers that compared with this method
9 later accepted papers report TransVG as a baseline in their results tables; 8 of them report a better number on most of the results they share with it.
- Towards Language-Guided Visual Recognition via Dynamic Convolutions (IJCV 2023 · better on 6 of 7 shared results)
- Context Disentangling and Prototype Inheriting for Robust Visual Grounding (TPAMI 2023 · better on 6 of 6 shared results)
- Dynamic MDETR: A Dynamic Multimodal Transformer Decoder for Visual Grounding (TPAMI 2023 · better on 1 of 3 shared results)
- DocEdit: Language-Guided Document Editing (AAAI 2023 · better on 2 of 2 shared results)
- Linking People across Text and Images Based on Social Relation Reasoning (AAAI 2023 · better on 1 of 1 shared result)
- Referring Expression Comprehension Using Language Adaptive Inference (AAAI 2023 · better on 1 of 1 shared result)
- SiRi: A Simple Selective Retraining Mechanism for Transformer-Based Visual Grounding (ECCV 2022 · better on 6 of 6 shared results)
- SeqTR: A Simple Yet Universal Network for Visual Grounding (ECCV 2022 · better on 4 of 5 shared results)
1 more not listed.
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.
- MAttNet — MAttNet: Modular Attention Network for Referring Expression Comprehension (CVPR 2018) · better on 1 of 1 shared result
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
- TransVG++: End-to-End Visual Grounding With Language Conditioned Vision Transformer (TPAMI 2023 · similarity 0.95)
- SimVG: A Simple Framework for Visual Grounding with Decoupled Multi-modal Fusion (NeurIPS 2024 · similarity 0.94)
- GeM-VG: Towards Generalized Multi-image Visual Grounding with Multimodal Large Language Models (AAAI 2026 · similarity 0.90)
- Referencing Where to Focus: Improving Visual Grounding with Referential Query (NeurIPS 2024 · similarity 0.90)
- Connecting the Dots: Training-Free Visual Grounding via Agentic Reasoning (AAAI 2026 · similarity 0.89)
- AugRefer: Advancing 3D Visual Grounding via Cross-Modal Augmentation and Spatial Relation-based Referring (AAAI 2025 · 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.
