Research map / Continual learning / Class-incremental
Class-Incremental Instance Segmentation via Multi-Teacher Networks
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 a Current Teacher Network to guide learning of new classes (ablation) | YOLACT incremental instance segmentation · PASCAL 2012 SBD test, new class (tvmonitor) | mAP (%): 33.38 -> 41.29 | Helped |
| Use a Current Teacher Network to guide new-class learning (ablation) | YOLACT incremental instance segmentation · PASCAL 2012 SBD test, all classes (20; add classes 11-20) | mAP (%): 35.49 -> 37.39 | Helped |
| Use a Current Teacher Network to guide new-class learning (ablation) | YOLACT incremental instance segmentation · PASCAL 2012 SBD test, new classes (11-20) | mAP (%): 32.29 -> 34.50 | Helped |
| Retain the Former Teacher Network to distill previous-class knowledge (ablation) | YOLACT incremental instance segmentation · PASCAL 2012 SBD test, all classes (20) | mAP (%): 1.77 -> 39.73 | Helped |
| Add instance distillation loss for prototypes and mask coefficients (ablation) | YOLACT incremental instance segmentation · PASCAL 2012 SBD test, all classes (20) | mAP (%): 1.88 -> 39.73 | Helped |
| Add bounding-box distillation loss (ablation) | YOLACT incremental instance segmentation · PASCAL 2012 SBD test, all classes (20) | mAP (%): 39.50 -> 39.73 | Helped |
| Add classification-score distillation loss (ablation) | YOLACT incremental instance segmentation · PASCAL 2012 SBD test, all classes (20) | mAP (%): 39.39 -> 39.73 | Helped |
| Use a Current Teacher Network for the final sequential class addition (ablation) | YOLACT incremental instance segmentation · PASCAL 2012 SBD test, all classes (20; sequential additions) | mAP (%): 33.76 -> 34.51 | Helped |
Closest papers by meaning
- Incrementer: Transformer for Class-Incremental Semantic Segmentation With Knowledge Distillation Focusing on Old Class (CVPR 2023 · similarity 0.88)
- Class-Incremental Learning by Knowledge Distillation With Adaptive Feature Consolidation (CVPR 2022 · similarity 0.87)
- Modeling the Background for Incremental Learning in Semantic Segmentation (CVPR 2020 · similarity 0.87)
- Decomposed Knowledge Distillation for Class-Incremental Semantic Segmentation (NeurIPS 2022 · similarity 0.87)
- Class Incremental Learning with Multi-Teacher Distillation (CVPR 2024 · similarity 0.85)
- CoMBO: Conflict Mitigation via Branched Optimization for Class Incremental Segmentation (CVPR 2025 · similarity 0.85)
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.