AI Paper Atlas

Research map / Detection and segmentation / Tracking

EgoTracks: A Long-term Egocentric Visual Object Tracking Dataset

Hao Tang, Kevin J Liang, Kristen Grauman, Matt Feiszli, Weiyao Wang

NeurIPS 2023 · Detection and segmentation › Tracking · approach: Transformer-based video object tracking · 6 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
finetune on VQ response-track subset (ablation)STARK · EgoTracks VCT-VSF-score: 30.48 -> 33.53Improved
finetune on full EgoTracks instead of VQ subset (ablation)STARK · EgoTracks VCT-VSF-score: 33.53 -> 38.2Improved
finetune on EgoTracksToMP · EgoTracks VCT-VSF-score: 20.95 -> 28.11Improved
finetune on EgoTracksSiam-RCNN · EgoTracks VCT-VSF-score: 35.38 -> 41.41Improved
add multiscale scale augmentation (0.5-1.5) (ablation)STARK ft on VQ · EgoTracks VCT-VSAO: 38.94 -> 48.44Improved
add multiscale scale augmentation (0.5-1.5) (ablation)STARK ft on VQ · EgoTracks VCT-VSF-score: 33.53 -> 41.92Improved
expand search size from 320 to 800 (ablation)STARK · EgoTracks VCT-VSF-score: 30.48 -> 43.74Improved
expand search size from 320 to 800 (ablation)STARK · EgoTracks VCT-VSAO: 35.99 -> 54.08Improved
increase context ratio from 2x to 3x (same SIS 320) (ablation)STARK · EgoTracks VCT-VSF-score: 33.53 -> 36.03Improved
increase context ratio from 2x to 4x (same SIS 320) (ablation)STARK · EgoTracks VCT-VSF-score: 33.53 -> 34.32No clear effect
reduce context ratio from 2x to 1x (same SIS 320) (ablation)STARK · EgoTracks VCT-VSF-score: 33.53 -> 26.81Hurt
initialize template adjacent to tracked frames (VCT-VC vs VCT-VS)STARK · EgoTracksF-score: 30.48 -> 34.02Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 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