AI Paper Atlas

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

SynCLIP: Synonym-Coherent Language-Image Pretraining for Robust Open-Vocabulary Dense Perception

Mingjie Xie, Guangjun He, Dongli Xu, Youtian Lin, Hongjue Li, Pengming Feng, Jian Guan, Yue Deng

CVPR 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 SSA and SAR together (ablation)F-ViT (DeCLIP baseline) · OV-COCOAP50 novel: 41.0 -> 43.6Improved
add SSA alone (ablation)F-ViT baseline · OV-COCOAP50 novel: 41.0 -> 41.1No clear effect
add SAR alone (ablation)F-ViT baseline · OV-COCOAP50 novel: 41.0 -> 42.2Improved
add SSA on top of SAR (ablation)SynCLIP · OV-COCOAP50 novel: 42.2 -> 43.6Improved
add synonyms as SSA reference text (ablation)SynCLIP · OV-COCOAP50 novel: 41.1 -> 42.1Improved
add definitions as SSA reference text (ablation)SynCLIP · OV-COCOAP50 novel: 41.1 -> 42.3Improved
use both synonyms and definitions (ablation)SynCLIP · OV-COCOAP50 novel: 41.1 -> 43.6Improved
balance alpha and beta at 0.5 instead of alpha 0.9, beta 0.1 (ablation)SAR aggregation · OV-COCOAP50 novel: 41.4 -> 43.6Improved
balance alpha and beta at 0.5 instead of alpha 0.1, beta 0.9 (ablation)SAR aggregation · OV-COCOAP50 novel: 41.7 -> 43.6Improved
increase top-k tokens from 1 to 7 (ablation)SAR token selection · OV-COCOAP50 novel: 42.5 -> 43.6Improved
increase top-k tokens from 7 to 20 (ablation)SAR token selection · OV-COCOAP50 novel: 43.6 -> 41.5Hurt
replace SynCLIP pretraining with CLIPSelf under synonym substitutionF-ViT · OV-COCO (synonym substitution)AP50 novel: 37.6 -> 28.8 (-8.8)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