AI Paper Atlas

Research map / Vision-language and multimodal / Human and embodied modeling

"“This Is My Unicorn, Fluffy”: Personalizing Frozen Vision-Language Representations"

Niv Cohen, Rinon Gal, Eli A. Meirom, Gal Chechik, Yuval Atzmon

ECCV 2022 · Vision-language and multimodal › Human and embodied modeling · approach: Language-grounded scene text and structure · 70 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
learn word embeddings in CLIP input space (PALAVRA) instead of AvgIm+TextCLIP · DeepFashion2 retrieval, rich queryMRR: 18.8 -> 28.4Improved
learn word embeddings in CLIP input space (PALAVRA) instead of AvgIm+TextCLIP · YTVOS retrieval, rich queryMRR: 47.2 -> 61.2Improved
use adapter on CLIP text output (COLLIE) with rich queries rather than concept-onlyCOLLIE · DeepFashion2 retrievalMRR: 13.8 -> 7.9Hurt
add test-time tuning of the concept code (ablation)PALAVRA · DeepFashion2 retrievalMRR: 22.1 -> 28.4Improved
add test-time tuning of the concept code (ablation)PALAVRA · YTVOS retrievalMRR: 47.1 -> 61.2Improved
add text augmentation with extended vocabulary (ablation)f_theta (no tuning) · DeepFashion2 retrieval testMRR: 19.1 -> 22.1Improved
add text augmentation with extended vocabulary (ablation)f_theta (no tuning) · YTVOS retrieval testMRR: 44.4 -> 47.1Improved
add cycle loss to ground-truth regularization (ablation)f_theta (no tuning) · DeepFashion2 retrieval testMRR: 19.2 -> 22.1Improved
add cycle loss to ground-truth regularization (ablation)f_theta (no tuning) · YTVOS retrieval testMRR: 41.4 -> 47.1Improved
add ground-truth regularization to cycle loss (ablation)f_theta (no tuning) · DeepFashion2 retrieval testMRR: 16.1 -> 22.1Improved
predict initial code with f_theta before tuning instead of random init (ablation)PALAVRA · YTVOS retrieval testMRR: 57.8 -> 61.2Improved
predict initial code with f_theta before tuning instead of random init (ablation)PALAVRA · DeepFashion2 retrieval testMRR: 27.5 -> 28.4Improved

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