Research map / Continual learning / Class-incremental
Incremental object detection with replay
Reduces forgetting in detectors using distillation and generative replay of old classes.
32 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2020–2026) · an approach in Class-incremental, keyword cluster “shot · incremental cil · feature”
State History Tried Checklist Preprints Who Export
Where it stands
Not enough benchmark data Fewer than four papers report the same benchmark, so there is no record to follow.
8 papers in 2023–24 and 17 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic grew, from 8.7% to 12.6%.
How this is decided
A benchmark is tracked when at least four papers of this approach report their own method on the same dataset and metric in at least two different years. Its record is the best number reported up to 2024 compared with the best reported in 2025–26: still improving means at least 1% better, flat means at most 0.2% better with at least two papers from 2025–26 trying. Numbers on a different scale from the rest (a fraction among percentages) and far outliers are left out. Results come from different papers with different setups, so they are signals, not head-to-head comparisons. The share of papers is counted at ICML, NeurIPS, ICLR, CVPR and AAAI.
How it got here
Most cited paper of each year
- 2020 · Learning to Segment the Tail (CVPR 2020 · 96 citations)
- 2021 · DER: Dynamically Expandable Representation for Class Incremental Learning (CVPR 2021 · 653 citations)
- 2022 · Overcoming Catastrophic Forgetting in Incremental Object Detection via Elastic Response Distillation (CVPR 2022 · 122 citations)
- 2023 · Continual Detection Transformer for Incremental Object Detection (CVPR 2023 · 93 citations)
- 2024 · SDDGR: Stable Diffusion-based Deep Generative Replay for Class Incremental Object Detection (CVPR 2024 · 42 citations)
- 2025 · Revisiting Generative Replay for Class Incremental Object Detection (CVPR 2025 · 7 citations)
- 2026 · YOLO-IOD: Towards Real Time Incremental Object Detection (AAAI 2026 · 4 citations)
Earliest papers here
- Learning to Segment the Tail (CVPR 2020 · 96 citations)
- DER: Dynamically Expandable Representation for Class Incremental Learning (CVPR 2021 · 653 citations)
- I3DOL: Incremental 3D Object Learning without Catastrophic Forgetting (AAAI 2021 · 25 citations)
What has been tried
19 papers here report 219 measured changes (a component added, removed or swapped, compared with a baseline): helped 75%, no clear effect 6%, hurt 14%, mixed 5%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Add pseudo-labeling | COCO 2017, 70+10 · AP (all categories) | 14.8 -> 38.6 | Helped | SDDGR: Stable Diffusion-based Deep Generative Replay for Class Incremental Object Detection (CVPR 2024 · 42 citations) |
| apply task-aware representation and selective inference instead of fine-tuning alone | Pascal VOC 2007 10+10 · Avg mAP@0.5 | 50.6 -> 70.6 | Helped | Learning Task-Aware Language-Image Representation for Class-Incremental Object Detection (AAAI 2024 · 12 citations) |
| Add IGR image-level generative replay | PASCAL VOC 2007 10-10 · mAP | 73.7 -> 75.1 | Helped | Revisiting Generative Replay for Class Incremental Object Detection (CVPR 2025 · 7 citations) |
| Add hierarchical distillation (Lk_hc) to Lk_ce + Lk_kd | HMDB51, 5 × 5 stages · accuracy (CNN) | 44.36 -> 47.34 | Helped | Hypercorrelation Evolution for Video Class-Incremental Learning (AAAI 2024 · 4 citations) |
| add CrossSDC to output distillation | MUSIC-21 · SDR | 6.32 -> 7.33 | Helped | Continual Audio-Visual Sound Separation (NeurIPS 2024 · 4 citations) |
| Add Decoupled Localization and Recognition | VOC 15+5 · All | 62.5 -> 65.7 | Helped | DCA: Dividing and Conquering Amnesia in Incremental Object Detection (AAAI 2025 · 1 citation) |
| replace LWF's keypoint-wise normalization with keypoint-oriented spatial distillation ℓKSD | Split MPII · AAA4 | 75.75 -> 76.93 | Helped | Incremental Object Keypoint Learning (CVPR 2025 · 0 citations) |
| add NSGP without prototype replay | PASCAL VOC 2007 5-5, 1-20 · mAP@0.5 | 58.4 -> 63.3 | Helped | Demystifying Catastrophic Forgetting in Two-Stage Incremental Object Detector (ICML 2025 · 0 citations) |
What reviewers will expect
Datasets and baselines named in the 6 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- PASCAL VOC 2007 · 2 of 6 papers
- MS COCO 2017 · 2 of 6 papers
Baselines
- Faster ILOD · 3 of 6 papers
- MMA · 3 of 6 papers
- OW-DETR · 3 of 6 papers
- PROB · 3 of 6 papers
- LwF · 2 of 6 papers
- Fine-tuning · 2 of 6 papers
- ORE · 2 of 6 papers
- ILOD-Meta · 2 of 6 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 25 papers; from OpenAlex and the papers' affiliations).
- Chinese Academy of Sciences · 2 papers
- University of Chinese Academy of Sciences · 2 papers
- Harbin Institute of Technology · 2 papers
- Tsinghua University · 2 papers
- Northwestern Polytechnical University · 2 papers
Use it
Download BibTeX of 31 key papersOpen the topic in the interactive map
Get alerts. Describe your idea in scime atlas and save it (free scime account): you get a weekly email when new preprints land close to it.
Related approaches
Same keyword cluster
- Exemplar-efficient class-incremental learning (111 papers)
- Class-balanced knowledge retention (73 papers)
- Few-shot class-incremental prototypes (47 papers)
Elsewhere in the topic
- Unknown-class mining for incremental segmentation (23 papers)
- Distillation-based continual segmentation (17 papers)
Benchmark numbers, ablations and baseline numbers are extracted by AI from each paper's full text and can be wrong; the paper is the authority. Data: results index of 2026-11-02, citation counts from OpenAlex (2026-10-09).