Research map / Vision-language and multimodal / Human and embodied modeling
Fine-Grained Semantically Aligned Vision-Language Pre-Training
Juncheng Li, XIN HE, Longhui Wei, Long Qian, Linchao Zhu, Lingxi Xie, Yueting Zhuang, Qi Tian, Siliang Tang
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 token-level Shapley interaction loss L_TSA to dual-encoder backbone (ablation) | LOUPE backbone · COCO detection (CC3M ablation) | mAP@0.3: 3.8 -> 7.6 | Improved |
| add L_TSA (ablation) | LOUPE backbone · MSCOCO I2T R@1 (CC3M ablation) | R@1: 31.0 -> 32.4 | Improved |
| add semantics-level Shapley interaction loss L_FSA (ablation) | LOUPE with L_TSA · COCO detection (CC3M ablation) | mAP@0.3: 7.6 -> 9.4 | Improved |
| add L_FSA (ablation) | LOUPE with L_TSA · MSCOCO T2I R@1 (CC3M ablation) | R@1: 26.2 -> 28.3 | Improved |
| add uncertainty-aware neural Shapley interaction learning (UNSIL) (ablation) | LOUPE · MSCOCO I2T R@1 (CC3M ablation) | R@1: 33.5 -> 33.3 | No clear effect |
| add UNSIL (ablation) | LOUPE · training time | sec/iter: 9.90 -> 1.93 | Improved |
| add UNSIL (ablation) | LOUPE · COCO detection (CC3M ablation) | mAP@0.3: 9.4 -> 9.0 | No clear effect |
| add UNSIL (ablation) | LOUPE · RefCOCO val (CC3M ablation) | accuracy: 4.1 -> 4.5 | No clear effect |
| LOUPE vs FILIP, with less pre-training data (240M vs 340M) | LOUPE · MSCOCO zero-shot T2I | R@1: 45.9 -> 50.1 | Improved |
| LOUPE vs CLIP on zero-shot classification | LOUPE · 11-dataset zero-shot classification | avg top-1 accuracy: 73.0 -> 76.1 | Improved |
| LOUPE vs CLIP on FER2013 and CIFAR10 | LOUPE · zero-shot classification | top-1 accuracy: 57.7 -> 53.3 (FER2013) | Mixed |
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.
- FILIP — FILIP: Fine-grained Interactive Language-Image Pre-Training (ICLR 2022) · better on 3 of 3 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
- Fine-Grained Verifiers: Preference Modeling as Next-token Prediction in Vision-Language Alignment (ICLR 2025 · similarity 0.89)
- Benchmarking Large Vision-Language Models on Fine-Grained Image Tasks: A Comprehensive Evaluation (ICLR 2026 · similarity 0.88)
- Progressive Multi-granular Alignments for Grounded Reasoning in Large Vision-Language Models (AAAI 2025 · similarity 0.88)
- GranViT: A Fine-Grained Vision Model With Autoregressive Perception For MLLMs (ICLR 2026 · similarity 0.88)
- Improving fine-grained understanding in image-text pre-training (ICML 2024 · similarity 0.87)
- Remodeling Semantic Relationships in Vision-Language Fine-Tuning (AAAI 2026 · 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.
