AI Paper Atlas

Research map / Continual learning / Class-incremental

Continual Semantic Segmentation via Structure Preserving and Projected Feature Alignment

Zihan Lin, Zilei Wang, Yixin Zhang

ECCV 2022 · Continual learning › Class-incremental · approach: Distillation-based continual segmentation · 24 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 pseudo labels for old classes to fine-tuning with cross-entropy (ablation)Fine-tuning · VOC 2012 overlapped 15-1mIoU (all classes): 4.5 -> 19.5Improved
add projected feature alignment L_ali to the full framework (ablation)SPPA · VOC 2012 overlapped 15-1mIoU (all classes): 53.8 -> 55.9Improved
add structure preserving loss L_str to the full framework (ablation)SPPA · VOC 2012 overlapped 15-1mIoU (all classes): 51.2 -> 55.9Improved
use projected feature alignment instead of pooled distillation (ablation)Lce + pseudo label baseline · VOC 2012 overlapped 15-1mIoU (all classes): 48.2 -> 50.5Improved
use projected feature alignment instead of pixel-wise feature alignment (ablation)Lce + pseudo label baseline · VOC 2012 overlapped 15-1mIoU (all classes): 33.7 -> 50.5Improved
add feature space constraints (pixel-wise alignment) to the pseudo-label baseline (ablation)Lce + pseudo label baseline · VOC 2012 overlapped 15-1mIoU (all classes): 18.9 -> 33.7Improved
use structure preserving loss instead of prototype alignment (ablation)SPPA · VOC 2012 overlapped 15-1old-class discriminability (t-SNE)Improved
SPPA versus PLOP with 10 classes per stepSPPA · ADE20K overlapped 100-10mIoU (all classes): 31.5 -> 31.5No clear effect
SPPA versus PLOP when adding five classes in one stepSPPA · VOC 2012 disjoint 15-5mIoU (all classes): 64.2 -> 69.0Improved

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