AI Paper Atlas

Research map / Detection and segmentation / CNN features

Leveraging Hidden Positives for Unsupervised Semantic Segmentation

Hyun Seok Seong, WonJun Moon, SuBeen Lee, Jae-Pil Heo

CVPR 2023 · Detection and segmentation › CNN features · approach: Context modeling for semantic segmentation · 43 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 hidden positives and local hidden positives to STEGODINO ViT-S/8 segmentation · COCO-stuffunsupervised accuracy: 48.3 -> 57.2Improved
Add hidden-positive trainingDINO ViT-B/8 segmentation · Potsdam-3unsupervised accuracy: 77.0 -> 82.4Improved
Add task-specific GHP and LHP together (ablation)HP · COCO-stuffunsupervised accuracy: 49.3 -> 57.2Improved
Add LHP (ablation)HP without task-specific GHP and LHP · COCO-stuffunsupervised accuracy: 49.3 -> 52.5Improved
Add task-specific GHP (ablation)HP without task-specific GHP and LHP · COCO-stuffunsupervised accuracy: 49.3 -> 55.0Improved
Use symmetric GHP assignment (ablation)HP · COCO-stuffunsupervised accuracy: 54.0 -> 57.2Improved
Add consistency regularizer (ablation)HP · COCO-stuffunsupervised accuracy: 55.8 -> 57.2Improved
Replace hidden-positive training with naive contrastive learning and photometric perturbations (ablation)DINO ViT-S/8 segmentation · COCO-stuffunsupervised accuracy: 37.8 -> 57.2Hurt
Replace gradient propagation with loss propagation (ablation)DINO ViT-S/16 HP · COCO-stuffunsupervised mIoU: 24.3 -> 23.2No clear effect
Replace gradient propagation with loss propagation (ablation)SelfPatch HP · COCO-stuffunsupervised accuracy: 56.1 -> 54.5No clear effect
Replace gradient propagation with loss propagation (ablation)DINO ViT-S/16 HP · COCO-stufflinear mIoU: 39.1 -> 40.5Improved
Replace gradient propagation with loss propagation (ablation)SelfPatch HP · COCO-stufflinear mIoU: 41.3 -> 41.4No clear effect

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