AI Paper Atlas

Research map / Detection and segmentation / Tracking

A New Dataset and a Distractor-Aware Architecture for Transparent Object Tracking

Alan Lukežič, Ziga Trojer, Jiří Matas, Matej Kristan

IJCV 2024 · Detection and segmentation › Tracking · approach: Memory and correlation-based MOT · 11 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
train on Trans2k in addition to opaque datasetsDiMP · TOTBAUC: improves by over 16%Improved
train on Trans2k with segmentation ground truthD3S · TOTBAUC: gains a remarkable 6%Improved
train on Trans2kSTARK · TOTBAUC: additional performance boost of 2.5%Improved
train on Trans2kTransATOM · TOTBAUC: gains 3.3%Improved
train ATOM on Trans2k instead of adding the TransATOM method modificationsATOM · TOTBAUC: outperforms this extension by 1.7%Improved
add fine-tuning on transparent objects (Trans2k) (ablation)DiTra · TOTBAUC: reduces the tracking performance by 4.5%Improved
add distractor-aware feature extraction branch (ablation)DiTra · TOTBAUC: causes a 5% performance dropImproved
add pose-aware feature extraction branch (ablation)DiTra · TOTBAUC: reduces the performance by 2%Improved
add most recent template in distractor-aware branch (ablation)DiTra · TOTBAUC: reduces the tracking performance by approximately 1%Improved
use DiTra instead of second-best STARK and TOMPDiTra · TOTBAUC: outperforms the second-best STARK and TOMP for approximately 5% in AUCImproved
use DiTra instead of STARK on opaque VOT2020DiTra · VOT2020EAO: outperforms the second-best STARK by 2% EAOImproved
use DiTra instead of MixFormer on opaque GoT10kDiTra · GoT10k testAUC: 73.2 -> 76.1Improved

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