Research map / Continual learning / Class-incremental
Bridging Non Co-occurrence with Unlabeled In-the-wild Data for Incremental 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 blind sampling and dual-teacher distillation with unlabeled in-the-wild data | Faster R-CNN incremental object detection · PASCAL VOC 2007 test, 19+1, no co-occurrence | mAP (all): 61.4 -> 72.2 | Helped |
| Use sampled unlabeled data without class overlap | Ours incremental object detector · PASCAL VOC 2007 test, 19+1, no co-occurrence | mAP (all): 61.4 -> 70.2 | Helped |
| Use Open Images as unlabeled data for blind sampling and distillation | Ours incremental object detector · PASCAL VOC 2007 test, 19+1, no co-occurrence | mAP (all): 61.4 -> 70.1 | Helped |
| Use the proposed method for 40+40 incremental detection | Faster R-CNN incremental object detection · MS COCO 2014 minival, no co-occurrence | AP: 22.5 vs 27.7 | Mixed |
| Add R-CNN distillation (ablation) | Standard Faster R-CNN losses with blind sampling · PASCAL VOC 2007 test, 19+1 | mAP (all): 62.9 -> 66.9 | Helped |
| Add image-level distillation with RoI masks (ablation) | Standard Faster R-CNN losses with blind sampling · PASCAL VOC 2007 test, 19+1 | mAP (all): 62.9 -> 67.1 | Helped |
| Add instance-level distillation with heatmaps (ablation) | Standard Faster R-CNN losses with blind sampling · PASCAL VOC 2007 test, 19+1 | mAP (all): 62.9 -> 66.2 | Helped |
| Add all three distillation losses (ablation) | Faster R-CNN student · PASCAL VOC 2007 test, 19+1 | mAP (all): 62.9 -> 68.7 | Helped |
| Add blind sampling strategy to the three distillation losses (ablation) | Faster R-CNN student · PASCAL VOC 2007 test, 19+1 | mAP (all): 68.7 -> 70.2 | Helped |
Closest papers by meaning
- Static-Dynamic Co-teaching for Class-Incremental 3D Object Detection (AAAI 2022 · similarity 0.91)
- Revisiting Generative Replay for Class Incremental Object Detection (CVPR 2025 · similarity 0.89)
- Demystifying Catastrophic Forgetting in Two-Stage Incremental Object Detector (ICML 2025 · similarity 0.89)
- DCA: Dividing and Conquering Amnesia in Incremental Object Detection (AAAI 2025 · similarity 0.88)
- Non-exemplar Domain Incremental Object Detection via Learning Domain Bias (AAAI 2024 · similarity 0.88)
- SDDGR: Stable Diffusion-based Deep Generative Replay for Class Incremental Object Detection (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.