AI Paper Atlas

Research map / Video / Video-language understanding

Hierarchical Event Memory for Accurate and Low-latency Online Video Temporal Grounding

Minghang Zheng, Yuxin Peng, Benyuan Sun, Yi Yang, Yang Liu

ICCV 2025 · Video › Video-language understanding · approach: Temporal video grounding · 7 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 hierarchical event memory in place of frame-level memory (ablation)OnVTG model · TACoSR1@IoU0.5: 34.88 -> 41.61Improved
add dynamic memory size configuration (ablation)OnVTG model with event memory · TACoSR1@IoU0.5: 41.61 -> 43.97Improved
add adaptive memory updating instead of FIFO (ablation)OnVTG model with event memory and dynamic size · TACoSR1@IoU0.5: 43.97 -> 44.19Improved
add future prediction branch (ablation)OnVTG model (full memory) · TACoSR1@IoU0.5: 44.19 -> 37.44Hurt
add future prediction branch (ablation)OnVTG model (full memory) · TACoSstart-time delay: 22.38s -> -1.28sImproved
add future prediction branch (ablation)OnVTG model · MADR50@IoU0.5: 29.21 -> 12.18Hurt
use event-based proposals instead of previous frame-level OnVTG baselineOurs w/o FP · TACoSR1@IoU0.5: 29.74 -> 44.19; +14.45%Improved
use hierarchical event memory instead of frame-level memory of HATOurs w/o FP · TACoSR1@IoU0.5: 34.15 -> 44.19; +10.04%Improved
use event-based OnVTG instead of previous baselineOurs w/o FP · MADR5@IoU0.3: 4.71 -> 15.76Improved
use event-based OnVTG instead of previous baselineOurs w/o FP · ActivityNet CaptionsR1@IoU0.5: 25.48 -> 45.29Improved
use hierarchical event memory model instead of previous baselineOurs · MADinference speed: 734.3 FPS -> 614.5 FPSHurt
allow future window to start before event start (a<0) (ablation)future prediction branch · TACoSaccuracy and delayMixed

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