Research map / Vision-language and multimodal / CLIP and zero-shot
Geometry-Preserving Unsupervised Alignment for Heterogeneous Foundation Models
Shuwen Yu, Zhanxuan Hu, Yi Zhao, Yonghang Tai, Huafeng 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 |
|---|---|---|---|
| add GPUA alignment with DINOv3 features | CLIP ViT-B/16 · Zero-shot classification avg of 11 | accuracy (%): 65.2 -> 77.4 | Improved |
| use 16 samples per class instead of full set | GPUA · Zero-shot classification avg of 11 | accuracy (%): 77.4 -> 76.5 | Hurt |
| add GPUA to CLIP on EuroSAT | CLIP · EuroSAT | accuracy (%): 48.3 -> 88.2 | Improved |
| add GPUA (DINOv3) to MaskCLIP | MaskCLIP · ADE20K | mIoU: 11.9 -> 15.9 | Improved |
| add GPUA (DINOv3) to MaskCLIP | MaskCLIP · PASCAL VOC20 | mIoU: 54.2 -> 65.7 | Improved |
| add GPUA (DINOv3) to SCLIP | SCLIP · PASCAL VOC20 | mIoU: 81.5 -> 87.8 | Improved |
| add GPUA (DINOv3) to SC-CLIP | SC-CLIP · ADE20K | mIoU: 20.1 -> 21.3 | Improved |
| add DINO features to CLIP features for correspondence mining (ablation) | GPUA · ImageNet | accuracy (%): 69.9 -> 74.3 | Improved |
| add CLIP features to DINO features for correspondence mining (ablation) | GPUA · UCF101 | accuracy (%): 75.3 -> 78.4 | Improved |
| add DINO features to CLIP features for correspondence mining (ablation) | GPUA · Pets | accuracy (%): 91.2 -> 94.5 | Improved |
| use THS loss instead of Triplet loss (ablation) | GPUA · EuroSAT | accuracy (%): 75.3 -> 80.3 | Improved |
| use THS loss instead of Adaptive hubness-aware loss (ablation) | GPUA · EuroSAT | accuracy (%): 73.2 -> 80.3 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- MaskCLIP — MaskCLIP: Masked Self-Distillation Advances Contrastive Language-Image Pretraining (CVPR 2023) · better on 2 of 2 shared results
- SCLIP — S-CLIP: Semi-supervised Vision-Language Learning using Few Specialist Captions (NeurIPS 2023) · better on 2 of 2 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
- Topology-Aware Representation Alignment for Semi-Supervised Vision-Language Learning (NeurIPS 2026 · similarity 0.89)
- UniDSeg: Unified Cross-Domain 3D Semantic Segmentation via Visual Foundation Models Prior (NeurIPS 2024 · similarity 0.88)
- G^3-LQ: Marrying Hyperbolic Alignment with Explicit Semantic-Geometric Modeling for 3D Visual Grounding (CVPR 2024 · similarity 0.87)
- DH-Set: Improving Vision-Language Alignment with Diverse and Hybrid Set-Embeddings Learning (CVPR 2025 · similarity 0.87)
- It's a (Blind) Match! Towards Vision-Language Correspondence without Parallel Data (CVPR 2025 · similarity 0.87)
- Mamba as a Bridge: Where Vision Foundation Models Meet Vision Language Models for Domain-Generalized Semantic Segmentation (CVPR 2025 · 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.
