Research map / Continual learning / Class-incremental
Learning at a Glance: Towards Interpretable Data-Limited Continual Semantic Segmentation via Semantic-Invariance Modelling
Bo Yuan, Danpei Zhao, Zhenwei Shi
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add semantic-invariant prototype matching (SPM) (ablation) | LAG (fine tuning base) · VOC 15-1 | mIoU all: 0.60 -> 62.21 | Improved |
| add sample-specific feature preserving to SPM (ablation) | LAG · VOC 15-1 | mIoU all: 62.21 -> 65.55 | Improved |
| add neuron-relevant semantic consistency (NSC) (ablation) | LAG · VOC 15-1 | mIoU all: 65.55 -> 66.08 | Improved |
| add NSC (ablation) | LAG · ADE 100-10 | mIoU all: 33.53 -> 33.56 | No clear effect |
| add NSC (ablation) | LAG · ISPRS 2-1 | mIoU all: 47.49 -> 48.67 | Improved |
| use uncertainty-aware pseudo-labelling instead of no threshold (ablation) | LAG · VOC 15-1 | mIoU: 59.71 -> 66.08 | Improved |
| use UPL instead of DCPL (ablation) | LAG · ADE 100-10 | mIoU: 33.68 -> 33.56 | Hurt |
| use UPL instead of fixed threshold (ablation) | LAG · ISPRS 2-1 | mIoU: 46.84 -> 48.67 | Improved |
| increase rho from 0 to 0.6 (ablation) | LAG · VOC 15-1 | mIoU: 57.48 -> 66.08 | Improved |
| set Gamma to 0.7 instead of 0 (ablation) | LAG · VOC 15-1 | mIoU: 49.37 -> 66.08 | Improved |
| swap ResNet-101 for Swin-T backbone (ablation) | LAG DeepLabv3 · ISPRS 2-1 | mIoU all: 48.67 -> 44.57 | Hurt |
| use DeepLabv3+ instead of DeepLabv3 (ablation) | LAG ResNet-101 · ISPRS 2-1 | mIoU all: 48.67 -> 50.07 | Improved |
Closest papers by meaning
- A Survey on Continual Semantic Segmentation: Theory, Challenge, Method and Application (TPAMI 2024 · similarity 0.93)
- Decoupling Continual Semantic Segmentation (AAAI 2026 · similarity 0.90)
- Beyond Background Shift: Rethinking Instance Replay in Continual Semantic Segmentation (CVPR 2025 · similarity 0.89)
- Towards Continual Universal Segmentation (CVPR 2025 · similarity 0.88)
- Deconfound Semantic Shift and Incompleteness in Incremental Few-shot Semantic Segmentation (AAAI 2025 · similarity 0.88)
- TIKP: Text-to-Image Knowledge Preservation for Continual Semantic Segmentation (AAAI 2024 · similarity 0.88)
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.
