Research map / Vision-language and multimodal / Human and embodied modeling
FG-CLIP 2: A Bilingual Fine-grained Vision-Language Alignment Model
Chunyu Xie, Bin Wang, Fanjing Kong, Jincheng Li, Dawei Liang, Ji Ao, Dawei Leng, Yuhui Yin
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 Textual Intra-modal Contrastive (TIC) loss (ablation) | FG-CLIP 2 ViT-B/16 · COCO80 bounding box classification | top-1 accuracy: 70.1 -> 74.9 | Improved |
| add TIC loss (ablation) | FG-CLIP 2 ViT-B/16 · FG-OVD Hard | top-1 accuracy: 51.6 -> 52.3 | Improved |
| add cross-modal rank loss with global threshold synchronization (ablation) | FG-CLIP 2 ViT-B/16 · FG-OVD Hard | top-1 accuracy: 50.9 -> 52.3 | Improved |
| add cross-modal rank loss (ablation) | FG-CLIP 2 ViT-B/16 · COCO80 bounding box classification | top-1 accuracy: 74.0 -> 74.9 | Improved |
| add both LCMR and LTIC (ablation) | FG-CLIP 2 ViT-B/16 · COCO80 bounding box classification | top-1 accuracy: 62.7 -> 74.9 | Improved |
| add LCMR and LTIC (ablation) | FG-CLIP 2 ViT-B/16 · ImageNet-V2 | top-1 accuracy: 72.0 -> 72.2 | No clear effect |
| add Chinese data in stage 2 (bilingual vs English-only) (ablation) | FG-CLIP 2 · Chinese and English benchmarks | retrieval/classification | Improved |
| replace SigLIP 2 with FG-CLIP 2 encoder in LLaVA-1.5 | LLaVA-1.5 · GQA | accuracy: 62.0 -> 64.0 | Improved |
| replace SigLIP 2 with FG-CLIP 2 in LLaVA-1.5 | LLaVA-1.5 · MMBench-CN Dev | accuracy: 66.2 -> 67.6 | No clear effect |
| add fine-grained region training on top of SigLIP 2 weights | FG-CLIP 2 ViT-B/16 · FG-OVD Hard | top-1 accuracy: 24.9 -> 52.3 | Improved |
| use FG-CLIP 2 as the alignment model in LLMDet open-vocabulary detection | LLMDet Swin-T · LVIS minival | AP: 47.9 -> 51.6 | Improved |
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.
- FG-CLIP — FG-CLIP: Fine-Grained Visual and Textual Alignment (ICML 2025) · better on 6 of 6 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
- FG-CLIP: Fine-Grained Visual and Textual Alignment (ICML 2025 · similarity 0.93)
- FineLIP: Extending CLIP's Reach via Fine-Grained Alignment with Longer Text Inputs (CVPR 2025 · similarity 0.91)
- Semantic and Expressive Variations in Image Captions Across Languages (CVPR 2025 · similarity 0.90)
- SmartCLIP: Modular Vision-language Alignment with Identification Guarantees (CVPR 2025 · similarity 0.90)
- Meta CLIP 2: A Worldwide Scaling Recipe (NeurIPS 2025 · similarity 0.89)
- Unlocking Fine-Grained Perception in CLIP via Structurally-Aware Latent Masked Modeling (NeurIPS 2026 · 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.
