Research map / Continual learning / Class-incremental
REAL-OW: Rehearsal-free Open World Object Detection with Low-Rank Adaptation and Dual-Stage Objectness Modeling
Huazhong Zhang, Xiaowen Fu, Yang Zhang, Linlin Shen, Jinbao Wang
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 L_AGS instead of standard distillation (ablation) | REAL-OW · M-OWODB Task 4 | mAP (both): 36.6 -> 38.7 | Improved |
| add L_AGS instead of standard distillation (ablation) | REAL-OW · M-OWODB Task 1 | U-Recall: 26.7 -> 25.5 | Hurt |
| add CG-NLL distance instead of Mahalanobis (ablation) | REAL-OW · M-OWODB Task 2 | U-Recall: 25.1 -> 26.9 | Improved |
| add learning phase (ablation) | REAL-OW · M-OWODB Task 2 | previously known mAP: 47.8 -> 59.4 | Improved |
| add consolidation phase (ablation) | REAL-OW · M-OWODB Task 2 | U-Recall: 21.0 -> 26.9 | Improved |
| add consolidation phase (ablation) | REAL-OW · M-OWODB Task 2 | mAP (both): 49.2 -> 49.0 | Hurt |
| add S_obj objectness score (ablation) | REAL-OW · M-OWODB Task 2 | mAP (both): 43.5 -> 49.0 | Improved |
| add L_KL (ablation) | REAL-OW · M-OWODB Task 2 | U-Recall: 20.4 -> 26.9 | Improved |
| add L_margin (ablation) | REAL-OW · M-OWODB Task 2 | U-Recall: 20.8 -> 26.9 | Improved |
| add L_LIO (ablation) | REAL-OW · M-OWODB Task 2 | mAP (both): 46.7 -> 49.0 | Improved |
| increase LoRA ranks beyond GA 16 / SA 8 (to GA 64 / SA 64) (ablation) | REAL-OW LoRA ranks · M-OWODB Task 3 | mAP: 42.23 -> 38.43 | Hurt |
| LoRA on q,k,v instead of q,v (ablation) | REAL-OW LoRA placement · M-OWODB Task 3 (R_GA:R_SA 64:32) | mAP: 40.22 -> 38.67 | Hurt |
Closest papers by meaning
- Open-World Objectness Modeling Unifies Novel Object Detection (CVPR 2025 · similarity 0.91)
- OW-OVD: Unified Open World and Open Vocabulary Object Detection (CVPR 2025 · similarity 0.89)
- Looking Beyond the Known: Towards a Data Discovery Guided Open-World Object Detection (NeurIPS 2025 · similarity 0.89)
- Hyp-OW: Exploiting Hierarchical Structure Learning with Hyperbolic Distance Enhances Open World Object Detection (AAAI 2024 · similarity 0.88)
- Detecting Open World Objects via Partial Attribute Assignment (CVPR 2025 · similarity 0.88)
- Exploring Orthogonality in Open World Object Detection (CVPR 2024 · similarity 0.88)
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.
