AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Efficient Diverse Ensemble for Discriminative Co-Tracking

Kourosh Meshgi, Shigeyuki Oba, Shin Ishii

CVPR 2018 · Label-efficient and robust learning › Recognition and long-tail · approach: Group-robust classification · 29 citations (OpenAlex)

Read the paper

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
combine uncertain-sample updates with artificially generated diversity data (ablation)DEDT ensemble · OTB50success AUCImproved
use artificial diversity data alone instead of combining it with tracker samples (ablation)DEDT ensemble · OTB50success AUCHurt
replace artificial diversity samples with nearest real image patches (ablation)DEDT · OTB50tracking performanceNo clear effect
query disputed labels from the auxiliary classifier rather than isolate it and fuse outputs afterward (ablation)DEDT · OTB50tracking performanceImproved
keep the auxiliary classifier fixed after training on the first frame (ablation)DEDT · OTB50tracking performanceHurt
update the auxiliary classifier every frame, removing its long-memory behavior (ablation)DEDT · OTB50tracking performanceHurt
increase the query threshold so the auxiliary classifier labels more, including slightly uncertain samples (ablation)DEDT · tracking performancetracking performanceHurt
use DEDT instead of CCOTvisual tracking · OTB50Avg. Succ: 0.75 -> 0.74Hurt
use DEDT instead of CCOTvisual tracking · OTB100Avg. Succ: 0.74 -> 0.69Hurt
use DEDT instead of SRDCFvisual tracking · VOT2015Accuracy: 0.56 -> 0.58Improved

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