AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Dual-stage Hyperspectral Image Classification Model with Spectral Supertoken

Peifu Liu, Tingfa Xu, Jie Wang, Huan Chen, Huiyan Bai, Jianan Li

ECCV 2024 · Label-efficient and robust learning › Recognition and long-tail · approach: Learning from unlabeled and positive-unlabeled data · 7 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 semantic features to supertoken clustering (ablation)DSTC-R · WHU-OHSCF1: 0.669 -> 0.717Improved
add first-order spectral derivative features (ablation)DSTC-R · WHU-OHSCF1: 0.717 -> 0.721Improved
add second-order spectral derivative features (ablation)DSTC-R · WHU-OHSCF1: 0.721 -> 0.713Hurt
use class-proportion soft labels instead of hard labels (ablation)DSTC-R · WHU-OHSCF1: 0.443 -> 0.721Improved
use soft labels on supertokens instead of dense pixel-level CE (ablation)DSTC-R · WHU-OHSCF1: 0.699 -> 0.721Improved
increase centroids per segment from 1 to 4 (ablation)DSTC-R · WHU-OHSCF1: 0.697 -> 0.721Improved
increase centroids per segment from 4 to 16 (ablation)DSTC-R · WHU-OHSCF1: 0.721 -> 0.685Hurt
increase crop factor F from 4 to 16 (ablation)DSTC-R · WHU-OHSCF1: 0.698 -> 0.721Improved
use DSTC dual-stage supertoken model versus prior methods on large datasetWHU-OHS · WHU-OHSCF1: 0.683 -> 0.723Improved
use DSTC on small patch-based datasets vs CVSSNIP, KSC, UP · IP / KSC / UPOA: IP 0.9827 -> 0.9875; KSC 0.9870 -> 0.9860; UP 0.9988 -> 0.9982Mixed

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