AI Paper Atlas

Research map / Label-efficient and robust learning / Semi-supervised learning

Two Losses, One Goal: Balancing Conflict Gradients for Semi-supervised Semantic Segmentation

Rui Sun, Huayu Mai, Wangkai Li, Yujia Chen, Yuan Wang

ICCV 2025 · Label-efficient and robust learning › Semi-supervised learning · approach: Consistency training for semi-supervised segmentation · 5 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 POS (Pareto weighting) without MEO (ablation)UniMatch V1 · Pascal 1/16 (92)mIoU: 75.2 -> 77.0Improved
add POS without MEO (ablation)UniMatch V1 · Pascal 1/8 (183)mIoU: 77.2 -> 78.1Improved
add POS without MEO (ablation)UniMatch V1 · COCO 1/512mIoU: 31.9 -> 33.4Improved
add POS without MEO (ablation)UniMatch V1 · COCO 1/256mIoU: 38.9 -> 39.8Improved
add MEO on top of POS (ablation)UniMatch V1 · Pascal 1/16 (92)mIoU: 77.0 -> 77.6Improved
add MEO on top of POS (ablation)UniMatch V1 · COCO 1/512mIoU: 33.4 -> 34.0Improved
add MEO on top of POS (ablation)UniMatch V1 · Pascal 1/8 (183)mIoU: 78.1 -> 78.7Improved
add MEO on top of POS (ablation)UniMatch V1 · COCO 1/256mIoU: 39.8 -> 40.3Improved
replace fixed weights (2:1) with dynamic POS weights (ablation)UniMatch V1 · Pascal 1/16 (92)mIoU: 74.9 -> 77.6Improved
replace fixed weights (1:2) with dynamic POS weights (ablation)UniMatch V1 · Pascal 1/16 (92)mIoU: 74.1 -> 77.6Improved
replace equal fixed weights with dynamic POS weights (ablation)UniMatch V1 · Pascal 1/16 (92)mIoU: 75.2 -> 77.6Improved
add POS+MEO to UniMatch V2 (DINOv2-S)UniMatch V2 · Pascal 1/16 (92)mIoU: 79.0 -> 80.7Improved

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