AI Paper Atlas

Research map / Video / Video-language understanding

Text-Based Temporal Localization of Novel Events

Sudipta Paul, Niluthpol Chowdhury Mithun, Amit K. Roy-Chowdhury

ECCV 2022 · Video › Video-language understanding · approach: Dense video captioning · 10 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
use semantically relevant seen moment as support instead of irrelevant one (ablation)TLRR · Charades-STA UnseenR@1 IoU@0.5: 20.30 -> 33.15Improved
use semantically relevant seen moment as support instead of random one (ablation)TLRR · Charades-STA UnseenR@1 IoU@0.5: 28.71 -> 33.15Improved
use relevant instead of irrelevant support (ablation)TLRR · Charades-STA UnseenmIoU: 22.48 -> 31.29Improved
use relevant instead of random support (ablation)TLRR · Charades-STA UnseenR@5 IoU@0.7: 40.24 -> 42.40Improved
add negative relational loss Lneg to Lintra (ablation)TLRR · Charades-STA Unseensupport relevance sensitivity: within 1% -> 5% -15% differenceImproved
replace 2D-TAN with TLRR2D-TAN · Charades-STA UnseenR@1 IoU@0.7: 13.33 -> 16.22Improved
replace 2D-TAN with TLRR2D-TAN · ActivityNet Captions UnseenR@5 IoU@0.7: 24.01 -> 36.66Improved
replace SCDM with TLRRSCDM · ActivityNet Captions UnseenR@5 IoU@0.5: 46.38 -> 53.31Improved
use TLRR instead of SCDM for seen-vs-unseen gapSCDM · Charades-STA Unseenavg gap: 19.80 -> 3.37Improved
use TLRR instead of 2D-TAN for seen-vs-unseen gap2D-TAN · ActivityNet Captions Unseenavg gap: 16.18 -> 7.13Improved
use TLRR instead of 2D-TAN on seen queries2D-TAN · Charades-STA Unseen (seen queries)R@1 IoU@0.5: 37.95 -> 34.83Mixed
use TLRR instead of SCDM on seen queriesSCDM · ActivityNet Captions Unseen (seen queries)R@1 IoU@0.5: 34.66 -> 27.46Hurt

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