AI Paper Atlas

Research map / Detection and segmentation / Open-set and few-shot detection

ATAS: Any-to-Any Self-Distillation for Enhanced Open-Vocabulary Dense Prediction

Juan Yeo, Soonwoo Cha, Jiwoo Song, Hyunbin Jin, Taesup Kim

ICCV 2025 · Detection and segmentation › Open-set and few-shot detection · approach: Open-vocabulary semantic segmentation · 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
add Global-to-Local Distillation to plain CLIP (ablation)ATAS · F-ViT detectionAP50: 31.40 -> 45.98Improved
add Local-to-Local Distillation (ablation)ATAS · Detection AP50AP50: 45.98 -> 45.39Hurt
add Global-to-Global Distillation (ablation)ATAS · MaskCLIP segmentationmIoU: 37.89 -> 39.34Improved
use contrastive objective instead of cosine in GLD (ablation)ATAS GLD · MaskCLIP segmentationmIoU: 36.98 -> 37.62Improved
add selective patch weighting with contrastive GLD (ablation)ATAS GLD · Segmentation and detectionmIoU / AP50: 37.62 -> 37.6 mIoU; 45.88 -> 45.98 AP50No clear effect
remove GGD loss (ablation)ATAS student · ImageNet accuracyaccuracy: 40.92 -> 67.25Improved
increase LLD weight from 0.01 to 1 (ablation)ATAS · Segmentation vs detectionmIoU / AP50: seg 39.11 -> 42.31, det 45.87 -> 44.57 (with GGD 0.01 vs 1)Mixed
set contrastive temperature to 1 versus 0.1 or 10 (ablation)ATAS · Detection AP50AP50: 45.11 -> 46.38 (tau 0.1 -> 1)Improved
train CLIP with ATAS instead of CLIPSelf for zero-shot MaskCLIPCLIP ViT-B · Zero-shot segmentation (MaskCLIP)average mIoU: 26.4 -> 31.2Improved
replace CLIP weights with ATAS in CAT-Seg v2 fine-tuningCAT-Seg v2 · Fine-tuned segmentation averagemIoU: 48.6 -> 49.1Improved
replace CLIPSelf weights in CAT-Seg v2CAT-Seg v2 · Fine-tuned segmentation averagemIoU: 48.6 -> 48.3Hurt
ATAS versus CLIPSelf in F-ViT detectionF-ViT · OV-COCOAP50 novel: 33.6 -> 37.2Improved

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