Research map / Detection and segmentation / Tracking
Video Event Extraction via Tracking Visual States of Arguments
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Track argument visual states using pixel changes, displacements, temporal aggregation, and object interactions | SlowFast for verb classification · VidSitu validation | F1@5: 17.87 -> 21.40 | Helped |
| Track argument visual states using pixel changes, displacements, and temporal aggregation | SlowFast for semantic role prediction · VidSitu validation | CIDEr: 44.49 ±2.30 -> 48.46 ±1.84 | Helped |
| Add bounding-box displacements to pixel-based Object State Embedding (ablation) | OSE-pixel + OME · VidSitu validation | F1@5: 21.24 -> 21.34 | Helped |
| Add bounding-box displacements to pixel-based Object State Embedding (ablation) | OSE-pixel + OME · VidSitu validation | CIDEr: 47.82 ±2.12 -> 48.46 ±1.84 | Helped |
| Add Object Interaction Embedding (ablation) | OSE-pixel/disp + OME for verb classification · VidSitu validation | F1@5: 21.34 -> 21.40 | Helped |
| Add Object Interaction Embedding (ablation) | OSE-pixel/disp + OME for semantic role prediction · VidSitu validation | CIDEr: 48.46 ±1.84 -> 47.16 ±1.71 | Hurt |
| Vary the maximum number of tracked objects from 8 to 2 or 4 (ablation) | Proposed model for verb classification · VidSitu validation | F1@5: 21.40; 21.53; 21.29 | No clear effect |
| Use LSTM rather than average pooling for state aggregation (ablation) | OME for verb classification · VidSitu validation | F1@5: 21.36 -> 21.30 | No clear effect |
Closest papers by meaning
- VidEvent: A Large Dataset for Understanding Dynamic Evolution of Events in Videos (AAAI 2025 · similarity 0.87)
- Beyond Grounding: Extracting Fine-Grained Event Hierarchies across Modalities (AAAI 2024 · similarity 0.86)
- E.T. Bench: Towards Open-Ended Event-Level Video-Language Understanding (NeurIPS 2024 · similarity 0.86)
- Event-Equalized Dense Video Captioning (CVPR 2025 · similarity 0.86)
- EventLens: Event-Structure Reinforcement Learning for Video Understanding (NeurIPS 2026 · similarity 0.85)
- Video Event Extraction with Multi-View Interaction Knowledge Distillation (AAAI 2024 · similarity 0.85)
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.