Research map / Continual learning / Class-incremental
Static-Dynamic Co-teaching for Class-Incremental 3D Object Detection
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 SDCoT instead of freeze-and-add or fine-tuning | VoteNet class-incremental 3D detection · SUN RGB-D val, |C_novel|=5 | mAP@0.25, all: 32.42 -> 57.21; 28.79 -> 57.21 | Helped |
| Use SDCoT instead of freeze-and-add or fine-tuning | VoteNet class-incremental 3D detection · ScanNet val, |C_novel|=4 | mAP@0.25, all: 39.47 -> 54.25; 14.05 -> 54.25 | Helped |
| Add mixed pseudo-label and ground-truth supervision for old and new classes | SDCoT · SUN RGB-D val, |C_novel|=5 | mAP@0.25, all: 28.79 -> 51.14 | Helped |
| Add static-teacher distillation loss (ablation) | SDCoT with mixed-label supervision · SUN RGB-D val, |C_novel|=5 | mAP@0.25, all: 51.14 -> 55.12 | Helped |
| Add dynamic-teacher consistency loss (ablation) | SDCoT with mixed-label supervision · SUN RGB-D val, |C_novel|=5 | mAP@0.25, novel: 50.12 -> 57.11 | Helped |
| Add both distillation and consistency losses (ablation) | SDCoT · SUN RGB-D val, |C_novel|=5 | mAP@0.25, all: 51.14 -> 57.21 | Helped |
| Add static-teacher distillation loss (ablation) | SDCoT · ScanNet val, |C_novel|=4 | mAP@0.25, all: 51.74 -> 53.19 | Helped |
| Add dynamic-teacher consistency loss (ablation) | SDCoT · ScanNet val, |C_novel|=4 | mAP@0.25, all: 51.74 -> 52.76 | Helped |
| Add center-aware distillation (ablation) | Classification-logit distillation in SDCoT · SUN RGB-D val, |C_novel|=5 | mAP@0.25, all: 57.21 -> 56.38 | Hurt |
| Add size-aware distillation (ablation) | Classification-logit distillation in SDCoT · SUN RGB-D val, |C_novel|=5 | mAP@0.25, all: 57.21 -> 56.98 | Hurt |
| Add center- and size-aware distillation (ablation) | Classification-logit distillation in SDCoT · SUN RGB-D val, |C_novel|=5 | mAP@0.25, all: 57.21 -> 56.45 | Hurt |
| Replace L2 distillation loss with cross-entropy or temperature-based knowledge distillation loss (ablation) | SDCoT · Class-incremental 3D object detection | mAP@0.25 | Hurt |
Closest papers by meaning
- Revisiting Generative Replay for Class Incremental Object Detection (CVPR 2025 · similarity 0.87)
- Learn How to See: Collaborative Embodied Learning for Object Detection and Camera Adjusting (AAAI 2024 · similarity 0.86)
- UNION: Unsupervised 3D Object Detection using Object Appearance-based Pseudo-Classes (NeurIPS 2024 · similarity 0.86)
- DCA: Dividing and Conquering Amnesia in Incremental Object Detection (AAAI 2025 · similarity 0.86)
- A-Teacher: Asymmetric Network for 3D Semi-Supervised Object Detection (CVPR 2024 · similarity 0.85)
- Demystifying Catastrophic Forgetting in Two-Stage Incremental Object Detector (ICML 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.