Research map / Continual learning / Class-incremental
Early Preparation Pays Off: New Classifier Pre-tuning for Class Incremental Semantic Segmentation
Zhengyuan Xie, Haiquan Lu, Jia-wen Xiao, Enguang Wang, Le Zhang, Xialei Liu
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 NeST new-classifier pre-tuning to MiB | MiB · Pascal VOC 2012 15-1 (ResNet-101) | mIoU (all): 38.2 -> 51.8 | Improved |
| add NeST to MiB in the hard 10-1 setting | MiB · Pascal VOC 2012 10-1 (ResNet-101) | mIoU (all): 10.2 -> 37.4 | Improved |
| add NeST to MiB with Swin-B | MiB · Pascal VOC 2012 15-1 (Swin-B) | mIoU (all): 36.9 -> 71.4 | Improved |
| add NeST to PLOP | PLOP · Pascal VOC 2012 15-1 (ResNet-101) | mIoU (all): 56.2 -> 63.1 | Improved |
| add NeST to RCIL | RCIL · Pascal VOC 2012 15-1 (ResNet-101) | mIoU (all): 58.9 -> 61.4 | Improved |
| add NeST to MiB in the 19-1 setting | MiB · Pascal VOC 2012 19-1 (ResNet-101) | mIoU (all): 69.6 -> 69.7 | No clear effect |
| add NeST to MiB on ADE20K 100-5 | MiB · ADE20K 100-5 (ResNet-101) | mIoU (all): 25.9 -> 32.7 | Improved |
| add NeST to MiB on ADE20K 100-50 | MiB · ADE20K 100-50 (ResNet-101) | mIoU (all): 34.9 -> 35.1 | No clear effect |
| use NeST transformation init instead of random classifier init (ablation) | MiB · Pascal VOC 2012 15-1 | mIoU (all): 34.1 -> 51.8 | Improved |
| tune background-initialized classifier directly (two-stage) vs NeST (ablation) | MiB · Pascal VOC 2012 15-1 | mIoU (all): 38.7 -> 51.8 | Improved |
| use designed initialization of importance and projection matrices instead of random (ablation) | NeST · Pascal VOC 2012 15-1 | mIoU (new classes 16-20): 7.5 -> 20.4 | Improved |
| add importance matrix on top of projection matrix (ablation) | NeST · Pascal VOC 2012 15-1 | mIoU (all): 44.3 -> 51.8 | Improved |
Closest papers by meaning
- Cs2K: Class-specific and Class-shared Knowledge Guidance for Incremental Semantic Segmentation (ECCV 2024 · similarity 0.90)
- Replay Without Saving: Prototype Derivation and Distribution Rebalance for Class-Incremental Semantic Segmentation (TPAMI 2025 · similarity 0.89)
- Preparing the Future for Continual Semantic Segmentation (ICCV 2023 · similarity 0.88)
- Adaptive Prototype Replay for Class Incremental Semantic Segmentation (AAAI 2025 · similarity 0.88)
- CoMBO: Conflict Mitigation via Branched Optimization for Class Incremental Segmentation (CVPR 2025 · similarity 0.87)
- Deconfound Semantic Shift and Incompleteness in Incremental Few-shot Semantic Segmentation (AAAI 2025 · similarity 0.86)
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.
