AI Paper Atlas

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

ICML 2026 · Vision-language and multimodal › CLIP and zero-shot · approach: Open-vocabulary alignment from VLMs · 0 citations (OpenAlex)

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 changedWhereMeasuredEffect
add GPUA alignment with DINOv3 featuresCLIP ViT-B/16 · Zero-shot classification avg of 11accuracy (%): 65.2 -> 77.4Improved
use 16 samples per class instead of full setGPUA · Zero-shot classification avg of 11accuracy (%): 77.4 -> 76.5Hurt
add GPUA to CLIP on EuroSATCLIP · EuroSATaccuracy (%): 48.3 -> 88.2Improved
add GPUA (DINOv3) to MaskCLIPMaskCLIP · ADE20KmIoU: 11.9 -> 15.9Improved
add GPUA (DINOv3) to MaskCLIPMaskCLIP · PASCAL VOC20mIoU: 54.2 -> 65.7Improved
add GPUA (DINOv3) to SCLIPSCLIP · PASCAL VOC20mIoU: 81.5 -> 87.8Improved
add GPUA (DINOv3) to SC-CLIPSC-CLIP · ADE20KmIoU: 20.1 -> 21.3Improved
add DINO features to CLIP features for correspondence mining (ablation)GPUA · ImageNetaccuracy (%): 69.9 -> 74.3Improved
add CLIP features to DINO features for correspondence mining (ablation)GPUA · UCF101accuracy (%): 75.3 -> 78.4Improved
add DINO features to CLIP features for correspondence mining (ablation)GPUA · Petsaccuracy (%): 91.2 -> 94.5Improved
use THS loss instead of Triplet loss (ablation)GPUA · EuroSATaccuracy (%): 75.3 -> 80.3Improved
use THS loss instead of Adaptive hubness-aware loss (ablation)GPUA · EuroSATaccuracy (%): 73.2 -> 80.3Improved

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.

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

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy