AI Paper Atlas

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

Alleviating Textual Reliance in Medical Language-guided Segmentation via Prototype-driven Semantic Approximation

Shuchang Ye, Usman Naseem, Mingyuan Meng, Jinman Kim

ICCV 2025 · Detection and segmentation › Open-set and few-shot detection · approach: Open-vocabulary semantic segmentation · 10 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 ProLearn prototype approximation instead of SGSeglanguage-guided segmentation · MosMedData+ 1% textDice: 0.3452 -> 0.7218Improved
use ProLearn instead of LViTlanguage-guided segmentation · MosMedData+ 1% textDice: 0.1677 -> 0.7218Improved
use ProLearn instead of GuideSeglanguage-guided segmentation · Kvasir-SEG 1% textDice: 0.5615 -> 0.8718Improved
use ProLearn instead of SGSeglanguage-guided segmentation · QaTa-COV19 50% textDice: 0.8641 -> 0.8667Improved
reduce paired text from 50% to 1% for ProLearnProLearn · QaTa-COV19Dice: 0.8667 -> 0.8566No clear effect
reduce paired text from 50% to 1% for SGSegSGSeg · MosMedData+Dice: 0.7455 -> 0.3452Hurt
reduce paired text from 50% to 1% for ProLearnProLearn · MosMedData+Dice: 0.7539 -> 0.7218Hurt
use ProLearn with 1% reports instead of Swin U-Net image-onlyimage-only segmentation · Kvasir-SEGDice: 0.398 -> 0.872Improved
use ProLearn with 1% reports instead of BiomedParseimage-only segmentation · Kvasir-SEGDice: 0.828 -> 0.872Improved
use prototype retrieval instead of LLM text generation (GPT-2)SGSeg-style text compensation · inference timetime: 136ms -> 4msImproved
use prototype retrieval instead of Llama3SGSeg-style text compensation · inference timetime: 1.2s -> 4msImproved
increase candidates k beyond a threshold (ablation)ProLearn · Dice sensitivityDiceHurt

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