AI Paper Atlas

Research map / Detection and segmentation / Tracking

Joint Group Feature Selection and Discriminative Filter Learning for Robust Visual Object Tracking

Tianyang Xu, Zhen-Hua Feng, Xiao-Jun Wu, Josef Kittler

ICCV 2019 · Detection and segmentation › Tracking · approach: Foundation-model and correlation-filter tracking · 233 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
add channel group selection (keep 5% of channels) to a DCF with deep features (ablation)DCF with ResNet50 features · OTB2015AUC: 55.49% -> 63.07%Improved
add spatial group selection to a DCF with hand-crafted features (ablation)DCF with CN, IC, HOG · OTB2015AUC: selection ratios 5%~40%Improved
add channel group selection to a DCF with hand-crafted features (ablation)DCF with CN, IC, HOG · OTB2015AUC: maintains performance when compressed to 60%No clear effect
add spatial group selection to a DCF with deep features (ablation)DCF with ResNet50 · OTB2015AUCNo clear effect
add channel selection term R_C (BaseLine_CS) (ablation)BaseLine DCF · OTB2013DP and AUC: DP by 10.1%, AUC by 10.6%Improved
add spatial selection term R_S (BaseLine_SS) (ablation)BaseLine DCF · OTB2013DP and AUC: DP by 6.1%, AUC by 6.7%Improved
add low-rank temporal smoothness term R_T (BaseLine_LR) (ablation)BaseLine DCF · OTB2013DP and AUCImproved
combine all three terms (BaseLine_ALL = GFS-DCF) (ablation)BaseLine DCF · OTB2013DP and AUCImproved
vary lambda1, lambda2 in [0.01, 100] and lambda3 in [10, 100] (ablation)GFS-DCF · OTB2015DP and AUC: stable for lambda1, lambda2 in [0.01, 100], lambda3 in [10, 100]No clear effect
use GFS-DCF instead of ECOOTB2015 tracking · OTB2015OP/CLE: by 4.1%/4.5 pixelsImproved
use GFS-DCF instead of C-COTOTB2015 tracking · OTB2015OP/CLE: by 6.7%/3.7 pixelsImproved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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