Research map / Vision-language and multimodal / CLIP and zero-shot
X2-VLM: All-in-One Pre-Trained Model for Vision-Language Tasks
Yan Zeng, Xinsong Zhang, Hang Li, Jiawei Wang, Jipeng Zhang, Wangchunshu Zhou
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 multi-grained aligning (ablation) | X2-VLM · Flickr30K TR | R@1: 96.6 -> 98.0 | Improved |
| add multi-grained aligning (ablation) | X2-VLM · OVAD | mAP: 23.1 -> 27.9 | Improved |
| add box prediction loss (ablation) | X2-VLM · RefCOCO+ testB | accuracy: 66.0 -> 76.7 | Improved |
| add box prediction loss (ablation) | X2-VLM · Flickr30K IR | R@1: 89.6 -> 89.0 | Hurt |
| add region data (ablation) | X2-VLM · OVAD | mAP: 22.3 -> 27.9 | Improved |
| add region data (ablation) | X2-VLM · RefCOCO+ testB | accuracy: 69.3 -> 76.7 | Improved |
| add object data (ablation) | X2-VLM · Flickr30K IR | R@1: 86.8 -> 89.0 | Improved |
| add both multi-grained aligning and box prediction (full X2-VLM) (ablation) | X2-VLM · RefCOCO+ testA | accuracy: 78.6 -> 88.6 | Improved |
| add both multi-grained aligning and box prediction (full X2-VLM) (ablation) | X2-VLM · OVAD | mAP: 20.6 -> 27.9 | Improved |
| use temporal attention instead of average pooling over frames (ablation) | X2-VLM · MSRVTT retrieval | IR R@1: 47.6 -> 45.6 | Hurt |
| use temporal attention instead of average pooling over frames (ablation) | X2-VLM · MSVD-QA | accuracy: 52.8 -> 52.1 | Hurt |
| replace text encoder with XLM-R, no multilingual pretraining | X2-VLM base · Flickr30K DE | avg recall: 93.3 -> 94.0 | Improved |
Later papers that compared with this method
1 later accepted paper reports X2VLM base as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Cross-Modal and Uni-Modal Soft-Label Alignment for Image-Text Retrieval (AAAI 2024 · better on 1 of 1 shared result)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- BLIP base — BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generation (ICML 2022) · better on 4 of 4 shared results
- BEiT-3 (1.9B) — BEiT: BERT Pre-Training of Image Transformers (ICLR 2022) · better on 1 of 1 shared result
- VLMo large — VLMo: Unified Vision-Language Pre-Training with Mixture-of-Modality-Experts (NeurIPS 2022) · 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
- Locality Alignment Improves Vision-Language Models (ICLR 2025 · similarity 0.91)
- Synthesize Diagnose and Optimize: Towards Fine-Grained Vision-Language Understanding (CVPR 2024 · similarity 0.91)
- Benchmarking Large Vision-Language Models on Fine-Grained Image Tasks: A Comprehensive Evaluation (ICLR 2026 · similarity 0.91)
- Kernel-based Unsupervised Embedding Alignment for Enhanced Visual Representation in Vision-language Models (ICML 2025 · similarity 0.90)
- Same or Not? Enhancing Visual Perception in Vision-Language Models (CVPR 2026 · similarity 0.90)
- VL-SAE: Interpreting and Enhancing Vision-Language Alignment with a Unified Concept Set (NeurIPS 2025 · similarity 0.90)
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.
