Research map / Continual learning / Class-incremental
Background Adaptation with Residual Modeling for Exemplar-Free Class-Incremental Semantic Segmentation
Anqi Zhang, Guangyu Gao
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 Background Adaptation residual channel instead of shared background classifier (ablation) | BAR baseline · Pascal VOC 15-1 | mIoU all: 61.1 -> 66.3 | Improved |
| add Background Feature Distillation (ablation) | BAR · Pascal VOC 15-1 | mIoU novel: 30.5 -> 37.1 | Improved |
| add Group Knowledge Distillation on top of BFD (ablation) | BAR · Pascal VOC 15-1 | mIoU all: 67.3 -> 67.6 | No clear effect |
| add triplet background adaptation loss BgA- (ablation) | BAR · Pascal VOC 15-1 | mIoU novel: 36.5 -> 43.4 | Improved |
| add BgA+ loss without BgA- (ablation) | BAR · Pascal VOC 15-1 | mIoU all: 67.6 -> 66.5 | Hurt |
| combine BgA+ and BgA- losses (ablation) | BAR · Pascal VOC 15-1 | mIoU all: 69.0 -> 70.0 | Improved |
| apply Filter operation that zeroes positive adaptation logits (ablation) | BAR · Pascal VOC 15-1 | mIoU all: 60.8 -> 70.0 | Improved |
| use Background Feature Distillation instead of MSE over all regions (ablation) | BAR · Pascal VOC 15-1 | mIoU novel: 44.3 -> 45.9 | Improved |
| use triplet loss for background region instead of BCE-1 (ablation) | BAR · Pascal VOC 15-1 | mIoU all: 69.6 -> 70.0 | Improved |
| use BAR vs DKD on short-term ADE20K tasks | BAR · ADE20K 100-50 | mIoU all: 36.0 -> 35.7 | Hurt |
Closest papers by meaning
- Mitigating Background Shift in Class-Incremental Semantic Segmentation (ECCV 2024 · similarity 0.93)
- Modeling the Background for Incremental and Weakly-Supervised Semantic Segmentation (TPAMI 2021 · similarity 0.91)
- Modeling the Background for Incremental Learning in Semantic Segmentation (CVPR 2020 · similarity 0.90)
- Beyond Background Shift: Rethinking Instance Replay in Continual Semantic Segmentation (CVPR 2025 · similarity 0.88)
- Adaptive Prototype Replay for Class Incremental Semantic Segmentation (AAAI 2025 · similarity 0.88)
- Deconfound Semantic Shift and Incompleteness in Incremental Few-shot Semantic Segmentation (AAAI 2025 · 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.
