AI Paper Atlas

Research map / Vision-language and multimodal / CLIP and zero-shot

Reevaluating the Intra-Modal Misalignment Hypothesis in CLIP

Jonas Herzog, Yue Wang

CVPR 2026 · Vision-language and multimodal › CLIP and zero-shot · approach: Fine-grained improvements to CLIP · 1 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
replace OTI image-to-text inversion with PCA projection onto ImageNet class-name prompt subspaceimage-image retrieval, CLIP ViT-B/32 · Retrieval, 13-dataset averagemAP: 42.9 -> 49.0Improved
replace OTI with PCA projectionimage-image retrieval, SigLIP ViT-B/16 · Retrieval, 13-dataset averagemAP: 60.0 -> 62.8Improved
add PCA projection to raw image embeddingsCLIP ViT-L/14 retrieval · Retrieval, 13-dataset averagemAP: 53.7 -> 61.3Improved
add PCA projection to raw image embeddingsSigLIP and SigLIP2 retrieval · Retrieval, 13-dataset averagemAP: 57.2 -> 62.8 (SigLIP); 58.6 -> 64.4 (SigLIP2)Improved
add PCA projection (information removal) for weather and time-of-day labels (ablation)CLIP ViT-L/14 retrieval · BDD100k weather / time of daymAP: -2.3 mAP (weather); -1.0 mAP (time of day)Hurt
add image-image self-supervised objective (SigLIP2 vs SigLIP) (ablation)SigLIP · Cosine similarity histograms, ImageNet valclass overlap and modality gapNo clear effect
add image-image objective (SigLIP2 vs SigLIP)SigLIP retrieval · Retrieval, 13-dataset averagemAP: +1.2 mAPImproved
add PCA projection to prototype classifier (ablation)CLIP ViT-B/16 image-only few-shot · 11 datasets, 16-shotaccuracy: 73.5 -> 77.5Improved
add PCA projection to LDA classifier (ablation)CLIP ViT-B/16 image-only few-shot · 11 datasets, 16-shotaccuracy: 79.5 -> 79.6No clear effect
add PCA projection to prototype classifier (ablation)CLIP ViT-B/16 1-shot · 11 datasets, 1-shotaccuracy: 43.4 -> 50.7Improved
avoid image-image similarity (Tip-X, CODER) instead of image-space baselinesCLIP RN50 few-shot adaptation · 11 datasets few-shotaccuracyNo clear effect
swap CLIP for image-only DINO modelimage-image classification Dogs vs Cats · Dogs vs Cats 1-shotaccuracy: 84.2 -> 76.2 (DINOv2)Hurt

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