Research map / Continual learning / Class-incremental
CoMFormer: Continual Learning in Semantic and Panoptic Segmentation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use CoMFormer instead of PLOP | ADE20K continual panoptic segmentation · 100-10 | PQ all: 26.1 -> 29.7 | Helped |
| Use CoMFormer instead of PLOP | ADE20K continual panoptic segmentation · 100-5 | PQ all: 24.0 -> 28.2 | Helped |
| Use CoMFormer instead of RCIL | ADE20K continual semantic segmentation · 100-50 | mIoU all: 34.5 -> 38.4 | Helped |
| Use CoMFormer instead of RCIL | ADE20K continual semantic segmentation · 100-10 | mIoU all: 32.1 -> 32.3 | Helped |
| Use CoMFormer instead of RCIL | ADE20K continual semantic segmentation · 100-5 | mIoU all: 29.6 -> 30.9 | Helped |
| Add adaptive distillation and pseudo-labeling (ablation) | CoMFormer panoptic segmentation · ADE20K 100-50 | PQ all: 8.6 -> 36.7 | Helped |
| Add adaptive distillation and pseudo-labeling (ablation) | CoMFormer panoptic segmentation · ADE20K 100-10 | PQ all: 1.0 -> 29.7 | Helped |
| Add pseudo-labeling to adaptive distillation (ablation) | CoMFormer panoptic segmentation · ADE20K 100-10 | PQ all: 18.4 -> 29.7 | Helped |
| Add adaptive distillation to pseudo-labeling (ablation) | CoMFormer panoptic segmentation · ADE20K 100-10 | PQ all: 25.8 -> 29.7 | Helped |
| Replace standard distillation with adaptive distillation (ablation) | CoMFormer panoptic segmentation · ADE20K 100-10 | PQ all: 26.0 -> 29.7 | Helped |
| Replace standard distillation with adaptive distillation (ablation) | CoMFormer panoptic segmentation · ADE20K 100-50 | PQ all: 36.4 -> 36.7 | Helped |
| Use CoMFormer instead of Mask2Former with the same continual-learning additions (ablation) | ADE20K continual panoptic segmentation · 100-10 | PQ all: 23.7 -> 29.7 | Helped |
Closest papers by meaning
- ECLIPSE: Efficient Continual Learning in Panoptic Segmentation with Visual Prompt Tuning (CVPR 2024 · similarity 0.91)
- Rethinking Query-based Transformer for Continual Image Segmentation (CVPR 2025 · similarity 0.91)
- Towards Continual Universal Segmentation (CVPR 2025 · similarity 0.90)
- Continual Segmentation with Disentangled Objectness Learning and Class Recognition (CVPR 2024 · similarity 0.89)
- Beyond Background Shift: Rethinking Instance Replay in Continual Semantic Segmentation (CVPR 2025 · similarity 0.88)
- SimCS: Simulation for Domain Incremental Online Continual Segmentation (AAAI 2024 · similarity 0.87)
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.