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
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add hierarchical event memory in place of frame-level memory (ablation) | OnVTG model · TACoS | R1@IoU0.5: 34.88 -> 41.61 | Improved |
| add dynamic memory size configuration (ablation) | OnVTG model with event memory · TACoS | R1@IoU0.5: 41.61 -> 43.97 | Improved |
| add adaptive memory updating instead of FIFO (ablation) | OnVTG model with event memory and dynamic size · TACoS | R1@IoU0.5: 43.97 -> 44.19 | Improved |
| add future prediction branch (ablation) | OnVTG model (full memory) · TACoS | R1@IoU0.5: 44.19 -> 37.44 | Hurt |
| add future prediction branch (ablation) | OnVTG model (full memory) · TACoS | start-time delay: 22.38s -> -1.28s | Improved |
| add future prediction branch (ablation) | OnVTG model · MAD | R50@IoU0.5: 29.21 -> 12.18 | Hurt |
| use event-based proposals instead of previous frame-level OnVTG baseline | Ours w/o FP · TACoS | R1@IoU0.5: 29.74 -> 44.19; +14.45% | Improved |
| use hierarchical event memory instead of frame-level memory of HAT | Ours w/o FP · TACoS | R1@IoU0.5: 34.15 -> 44.19; +10.04% | Improved |
| use event-based OnVTG instead of previous baseline | Ours w/o FP · MAD | R5@IoU0.3: 4.71 -> 15.76 | Improved |
| use event-based OnVTG instead of previous baseline | Ours w/o FP · ActivityNet Captions | R1@IoU0.5: 25.48 -> 45.29 | Improved |
| use hierarchical event memory model instead of previous baseline | Ours · MAD | inference speed: 734.3 FPS -> 614.5 FPS | Hurt |
| allow future window to start before event start (a<0) (ablation) | future prediction branch · TACoS | accuracy and delay | Mixed |
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.
- MiniROAD — MiniROAD: Minimal RNN Framework for Online Action Detection (ICCV 2023) · better on 3 of 3 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
- Training-free Video Temporal Grounding using Large-scale Pre-trained Models (ECCV 2024 · similarity 0.89)
- TimeExpert: An Expert-Guided Video LLM for Video Temporal Grounding (ICCV 2025 · similarity 0.89)
- TRACE: Temporal Grounding Video LLM via Causal Event Modeling (ICLR 2025 · similarity 0.88)
- VTG-LLM: Integrating Timestamp Knowledge into Video LLMs for Enhanced Video Temporal Grounding (AAAI 2025 · similarity 0.88)
- DeCafNet: Delegate and Conquer for Efficient Temporal Grounding in Long Videos (CVPR 2025 · similarity 0.86)
- ReVisionLLM: Recursive Vision-Language Model for Temporal Grounding in Hour-Long Videos (CVPR 2025 · similarity 0.86)
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.
