AI Paper Atlas

Research map / Continual learning / Class-incremental

Interference-Isolated Elastic Weight Consolidation and Knowledge Calibration for Incremental Object Detection

De Cheng, Mingyue Zeng, Zhipeng Xu, Di Xu, Nannan Wang, Xinbo Gao

ICLR 2026 · Continual learning › Class-incremental · approach: Incremental object detection with replay

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 changedWhereMeasuredEffect
add IKI-EWC to pseudo-labeling baseline (ablation)pseudo-labeling baseline · VOC 2007 (10-10)mAP@0.5 (1-20): 73.8 -> 75.1Improved
add IKI-EWC to pseudo-labeling baseline (ablation)pseudo-labeling baseline · VOC 2007 (10-5)mAP@0.5 (1-20): 66.4 -> 70.9Improved
add IKI-EWC to pseudo-labeling baseline (ablation)pseudo-labeling baseline · VOC 2007 (5-5)mAP@0.5 (1-20): 61.0 -> 65.0Improved
add PKC to pseudo-labeling baseline (ablation)pseudo-labeling baseline · VOC 2007 (10-10)mAP@0.5 (1-20): 73.8 -> 74.3Improved
add PKC to pseudo-labeling baseline (ablation)pseudo-labeling baseline · VOC 2007 (5-5)mAP@0.5 (1-20): 61.0 -> 61.4Improved
add PKC on top of IKI-EWC (ablation)IKI-EWC · VOC 2007 (10-10)mAP@0.5 (1-20): 75.1 -> 75.7Improved
IKI-EWC instead of EWC++ (ablation)EWC++ · VOC 2007 (10-10)mAP@0.5 (1-20): 74.5 -> 75.1Improved
IKI-EWC instead of EWC++ (ablation)EWC++ · VOC 2007 (10-5)mAP@0.5 (1-20): 69.4 -> 70.9Improved
IKI-EWC instead of uniform L2++ penalty (ablation)L2++ · VOC 2007 (10-10)mAP@0.5 (1-20): 58.8 -> 75.1Improved
add projection layer to PKC (vs PKC-NPC) (ablation)PKC · VOC 2007 (10-10)mAP@0.5 (1-20): 75.3 -> 75.7Improved
add projection layer to PKC (vs PKC-NPC) (ablation)PKC · VOC 2007 (5-5)mAP@0.5 (1-20): 64.3 -> 65.4Improved
increase regularization strength lambda from 20 to 100 (ablation)IKI-EWC · VOC 2007 (10-10)mAP@0.5 (1-20): 75.1 -> 73.9Hurt

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy