AI Paper Atlas

Research map / Continual learning / Class-incremental

Continual Learning and Unknown Object Discovery in 3D Scenes via Self-Distillation

Mohamed El Amine Boudjoghra, Jean Lahoud, Salman Khan, Hisham Cholakkal, Rao M Anwer, Fahad Shahbaz Khan

ECCV 2024 · Continual learning › Class-incremental · approach: Incremental object detection with replay · 0 citations (OpenAlex)

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
replace exemplar replay and student auto-labeling with exemplar-free self-distillation from the previous-task model3D-OWIS · ScanNet200 Split C Task T2U-Recall: 15.80 -> 41.31Improved
replace exemplar replay with self-distillation3D-OWIS · ScanNet200 Split C Task T2mAP previously known: 13.90 -> 26.02Improved
self-distillation versus 3D-OWIS on wilderness impactOpenDistill3D · ScanNet200 Split A Task T2WI: 0.007 -> 0.352Hurt
add Unknown-Known Separation (UKS) to naive self-distillation (ablation)OpenDistill3D · ScanNet200 Split C Task T2mAP all classes: 21.83 -> 23.30Improved
add Unknown-Known Separation (UKS) (ablation)OpenDistill3D · ScanNet200 Split C Task T2U-Recall: 36.57 -> 39.53Improved
add objectness weight function WI to cross-entropy loss (ablation)OpenDistill3D · ScanNet200 Split C Task T2mAP all classes: 23.30 -> 24.24Improved
use objectness score itself as weight instead of WI (ablation)OpenDistill3D · ScanNet200 Split C Task T1mAP all classes: 24.90 -> 22.08Hurt
use objectness score itself as weight instead of WI (ablation)OpenDistill3D · ScanNet200 Split A Task T1mAP all classes: 38.81 -> 39.09Mixed
add unknown-object self-distillation to closed-set Mask3D trainingMask3D · ScanNet200 val (w/o DBSCAN)mAP: 25.73 -> 27.37Improved
use OpenDistill3D as proposal network for OpenMask3DOpenMask3D · ReplicamAP50: 15.90 -> 18.40Improved

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