AI Paper Atlas

Research map / Video / Action recognition

A Sliding Window Scheme for Online Temporal Action Localization

Young Hwi Kim, Hyolim Kang, Seon Joo Kim

ECCV 2022 · Video › Action recognition · approach: Weakly supervised temporal action localization · 13 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 anchor-based OAT-OSN instead of OAD-grouping CAG-QILOn-TAL · THUMOS'14 (tIoU 0.5)mAP (%): 33.0 -> 47.1Improved
use anchor-based OAT-OSN instead of OAD-grouping CAG-QILOn-TAL · MUSES (tIoU 0.5)mAP (%): 4.2 -> 14.2Improved
use anchor-based OAT-OSN instead of OAD-grouping CAG-QILOn-TAL · BBDB (tIoU 0.5)mAP (%): 32.9 -> 63.4Improved
add the Online Suppression Network instead of naive thresholding (ablation)OAT · THUMOS'14 (tIoU 0.5)mAP (%): 43.0 -> 47.1Improved
add the Online Suppression Network instead of naive thresholding (ablation)OAT · BBDB (tIoU 0.5)mAP (%): 49.5 -> 63.4Improved
add the Online Suppression Network instead of naive thresholding (ablation)OAT · THUMOS'14 (tIoU 0.5)AEDT (seconds): -1.68 -> -1.25Hurt
add the Online Suppression Network instead of naive thresholding (ablation)OAT · MUSES (tIoU 0.5)AEDT (seconds): -13.31 -> -13.99Mixed
switch online OSN to video-level NMS post-processing (ablation)OAT · THUMOS'14 (tIoU 0.5)mAP (%): 47.1 -> 53.9Improved
switch online OSN to video-level NMS post-processing (ablation)OAT · MUSES (tIoU 0.5)mAP (%): 14.2 -> 19.9Improved
replace average pooling anchor encoding with 2-layer convnets (ablation)sliding window anchor model · THUMOS'14 (tIoU 0.5)mAP (%): 42.7 -> 47.4Improved
replace convnet anchor encoding with transformer encoder/decoder (ablation)sliding window anchor model · THUMOS'14 (tIoU 0.5)mAP (%): 47.4 -> 53.9Improved
raise naive confidence threshold from 0.1 to 0.4 (ablation)OAT-Naive · THUMOS'14average mAP: 36.4 -> 39.4Improved

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