AI Paper Atlas

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

Harnessing Vision Foundation Models for High-Performance, Training-Free Open Vocabulary Segmentation

Yuheng Shi, Minjing Dong, Chang Xu

ICCV 2025 · Detection and segmentation › Open-set and few-shot detection · approach: Open-vocabulary semantic segmentation · 34 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
use splice-then-segment instead of segment-then-splice (cosine correlation) (ablation)MaskCLIP · 8-benchmark averagemIoU: 34.1 -> 40.6Improved
use SAM attention weights instead of cosine similarity (ablation)MaskCLIP splice-then-segment · 8-benchmark averagemIoU: 40.6 -> 42.7Improved
use proposed affinity matrix instead of attention (ablation)MaskCLIP splice-then-segment · 8-benchmark averagemIoU: 42.7 -> 43.1Improved
add splice-then-segment with affinity matrix (ablation)ProxyCLIP (CLIP-B/16) · 8-benchmark averagemIoU: 41.0 -> 44.3Improved
add splice-then-segment with affinity matrix (ablation)ProxyCLIP (OpenCLIP-H/14) · 8-benchmark averagemIoU: 43.3 -> 47.1Improved
add SAM refinement (ablation)Trident (CLIP-B/16) · 8-benchmark averagemIoU: 44.3 -> 45.8Improved
add SAM refinement (ablation)Trident (OpenCLIP-H/14) · 8-benchmark averagemIoU: 47.1 -> 48.6Improved
add PAMR post-processing (ablation)Trident (CLIP-B/16) · 8-benchmark averagemIoU: 45.8 -> 43.9Hurt
use all three prompt types (mask, box, point) in SAM refinement (ablation)Trident · VOC21mIoU: 58.3 -> 67.1Improved
scale mask prompt with alpha=0.005 instead of 1.0 (ablation)Trident SAM refinement · VOC20mIoU: 78.7 -> 84.5Improved
increase input shorter side from 336 to 576 (ablation)ProxyCLIP · VOC20mIoU: 79.7 -> 73.4Hurt
increase input shorter side from 448-336-224 to 576-336-224 (ablation)Trident w/o SAM refine · Context59mIoU: 37.0 -> 37.2Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 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