Research map / Continual learning / Class-incremental
Representation Compensation Networks for Continual Semantic Segmentation
Check 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 representation compensation with a frozen merged branch and a trainable branch (ablation) | MiB · PASCAL VOC 2012 15-1 overlapped | mIoU (%): 36.1 -> 43.0 | Helped |
| Add branch merging to parallel convolutions (ablation) | Parallel-Conv · PASCAL VOC 2012 15-1 | mIoU (%): 40.1 -> 42.0 | Helped |
| Freeze the merged branch to retain previous-step knowledge (ablation) | Parallel-Conv + Merge · PASCAL VOC 2012 15-1 | mIoU (%): 42.0 -> 42.8 | Helped |
| Add drop-path branch aggregation (ablation) | RC with merged and frozen branches · PASCAL VOC 2012 15-1 | mIoU (%): 42.8 -> 43.0 | Helped |
| Add spatial pooled distillation (ablation) | MiB + RC · PASCAL VOC 2012 15-1 overlapped | mIoU (%): 43.0 -> 58.3 | Helped |
| Add channel pooled distillation (ablation) | MiB + RC · PASCAL VOC 2012 15-1 overlapped | mIoU (%): 43.0 -> 58.4 | Helped |
| Combine spatial and channel pooled distillation (ablation) | MiB + RC · PASCAL VOC 2012 15-1 overlapped | mIoU (%): 58.3 -> 59.4; 58.4 -> 59.4 | Helped |
| Use pooled cube distillation without representation compensation (ablation) | MiB · PASCAL VOC 2012 15-1 overlapped | mIoU (%): 36.1 -> 57.8 | Helped |
| Replace average pooling with strip pooling (ablation) | PLOP distillation framework · PASCAL VOC 2012 15-1 overlapped | mIoU (%): 56.1 -> 54.6 | Hurt |
| Use global average pooling instead of the tested average-pooling method (ablation) | PLOP distillation framework · PASCAL VOC 2012 15-1 overlapped | mIoU (%): 56.1 -> 36.1 | Hurt |
| Use max pooling instead of average pooling (ablation) | PLOP distillation framework · PASCAL VOC 2012 15-1 overlapped | mIoU (%): 56.1 -> 48.0 | Hurt |
| Use RC networks rather than PLOP | Continual class segmentation · PASCAL VOC 2012 15-1 disjoint | mIoU (%): 46.5 -> 54.7 | Helped |
Closest papers by meaning
- Beyond Background Shift: Rethinking Instance Replay in Continual Semantic Segmentation (CVPR 2025 · similarity 0.88)
- Towards Continual Universal Segmentation (CVPR 2025 · similarity 0.88)
- Adaptive Prototype Replay for Class Incremental Semantic Segmentation (AAAI 2025 · similarity 0.88)
- TIKP: Text-to-Image Knowledge Preservation for Continual Semantic Segmentation (AAAI 2024 · similarity 0.87)
- Enhancing Visual Continual Learning with Language-Guided Supervision (CVPR 2024 · similarity 0.86)
- ECLIPSE: Efficient Continual Learning in Panoptic Segmentation with Visual Prompt Tuning (CVPR 2024 · 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.