Research map / Video / Video segmentation and tracking
Online Generic Event Boundary Detection
Hyungrok Jung, Daneul Kim, Seunggyun Lim, Jeany Son, Jonghyun Choi
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 EST loss to transformer decoder baseline (ablation) | CEA baseline · Kinetics-GEBD | Avg F1: 0.607 -> 0.698 | Improved |
| add REST loss to transformer decoder baseline (ablation) | CEA baseline · Kinetics-GEBD | Avg F1: 0.607 -> 0.654 | Improved |
| add REST loss on top of EST loss (ablation) | EST-trained CEA · Kinetics-GEBD | Avg F1: 0.698 -> 0.691 | Hurt |
| add OBD to EST-only model (ablation) | EST-trained CEA · Kinetics-GEBD | Avg F1: 0.698 -> 0.659 | Hurt |
| add OBD to REST-only model (ablation) | REST-trained CEA · Kinetics-GEBD | Avg F1: 0.654 -> 0.692 | Improved |
| add OBD to EST+REST CEA (ablation) | ESTimator · Kinetics-GEBD | Avg F1: 0.691 -> 0.748 | Improved |
| use cosine distance instead of L1/L2/KL error (ablation) | ESTimator · Kinetics-GEBD | Avg F1: 0.733 -> 0.748 | Improved |
| keep boundary errors in OBD queue (vs inliers only) (ablation) | OBD · Kinetics-GEBD | Avg F1: 0.663 -> 0.748 | Improved |
| add batch-wise loss weighting (ablation) | ESTimator · Kinetics-GEBD | Avg F1: 0.743 -> 0.748 | Improved |
| input length 8 instead of 4 (ablation) | ESTimator · Kinetics-GEBD | Avg F1: 0.728 -> 0.748 | Improved |
| input length 16 or 32 instead of 8 (ablation) | ESTimator · Kinetics-GEBD | Avg F1: 0.742 and 0.745 vs 0.748 | No clear effect |
| increase REST region K from 3 to 11/13 (ablation) | ESTimator · Kinetics-GEBD | Avg F1: 0.756 at K=11 or 13 | Improved |
Closest papers by meaning
- Generic Event Boundary Detection: A Benchmark for Event Segmentation (ICCV 2021 · similarity 0.94)
- Fine-grained Dynamic Network for Generic Event Boundary Detection (ECCV 2024 · similarity 0.94)
- Generic Event Boundary Detection via Denoising Diffusion (ICCV 2025 · similarity 0.92)
- Streaming Detection of Queried Event Start (NeurIPS 2024 · similarity 0.83)
- FlexEvent: Towards Flexible Event-Frame Object Detection at Varying Operational Frequencies (NeurIPS 2025 · similarity 0.83)
- OnlineTAS: An Online Baseline for Temporal Action Segmentation (NeurIPS 2024 · similarity 0.83)
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.
