AI Paper Atlas

Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning

Uncertainty-Aware Learning for Zero-Shot Semantic Segmentation

Ping Hu, Stan Sclaroff, Kate Saenko

NeurIPS 2020 · Label-efficient and robust learning › Zero-/few-shot and meta-learning · approach: Generative feature synthesis for zero-shot · 35 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 uncertainty-aware losses (image-level and pixel-level) (ablation)ZSS baseline · Pascal-Context, 30 unseenunseen mIoU: 13.9 -> 18.6Improved
add uncertainty-aware losses (image-level and pixel-level) (ablation)ZSS baseline · Pascal-Context, 156 unseenunseen mIoU: 11.7 -> 13.3Improved
add uncertainty-aware losses (image-level and pixel-level) (ablation)ZSS baseline · ADE20K, 75 unseenunseen mIoU: 13.5 -> 15.2Improved
add a single uncertainty-aware loss branch (image-level or pixel-level alone) (ablation)ZSS baseline · Pascal-Context 156 / ADE20K 75 unseenunseen mIoU: -0.2 to +2.2Mixed
add uncertainty-aware learningZSS baseline · Pascal-Context, 10 unseenunseen mIoU: 27.9 -> 35.4Improved
add uncertainty-aware learningZSS baseline · ADE20K, 25 unseenoverall mIoU: 32.0 -> 31.9No clear effect
add uncertainty-aware learningZSS baseline · ADE20K, 50 unseenunseen mIoU: 10.1 -> 14.4Improved
use PSPNet backbone or FastText embeddings instead of DeepLabV3+/word2vec (ablation)uncertainty-aware ZSS · PC-30, PC-156, ADE-75unseen mIoUImproved
exclude training images containing unseen classes (ablation)baseline and uncertainty model · PC-30 vs PC-156, ADE-75unseen mIoUMixed

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