Research map / Video / Action recognition
Synchronization of Multiple Videos
Avihai Naaman, Ron Shapira Weber, Oren Freifeld
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 decoder (reconstruction) to the encoder-only 1D bottleneck (ablation) | TPL · Penn Action | CBC: 56.5% -> 97.3% | Improved |
| add decoder (reconstruction) to the encoder-only 1D bottleneck (ablation) | TPL · Penn Action | PLP: 51.3% -> 96.7% | Improved |
| enforce median-length prototype in ICAE (ablation) | TPL · Penn Action | CBC: 97.3% -> 100% | Improved |
| enforce median-length prototype in ICAE (ablation) | TPL · Penn Action | PLP: 96.7% -> 100% | Improved |
| add 1D bottleneck (full TPL) over alignment network without bottleneck (ablation) | TPL · Penn Action | CBC: 80.4% -> 100% | Improved |
| use alignment network without bottleneck instead of Euclidean baseline (ablation) | TPL · Penn Action | CBC: 64.6% -> 80.4% | Improved |
| replace DTAN ICAE with TPL on DINO-ViT features | alignment method · Penn Action | CBC: 0.572 -> 0.788 | Improved |
| replace DBA with TPL on DINO-ViT features | alignment method · Penn Action | total runtime (s): 5415 -> 534 | Improved |
| replace DBA with TPL on DINO-ViT features | alignment method · Penn Action | CBC: 0.756 -> 0.788 | Improved |
| replace DBA with TPL on CARL features | alignment method · Penn Action | CBC: 0.947 -> 0.962 | Improved |
| replace SoftDBA with TPL on OpenCLIP features, GenAI-MVS | alignment method · GenAI-MVS | CBC: 0.926 -> 0.942 | Improved |
| replace unsynchronized 1-NN retrieval with synchronized retrieval | frame retrieval · Penn Action | time (s): 30.1 -> 0.24 | Improved |
Closest papers by meaning
- Temporal Cycle-Consistency Learning (CVPR 2019 · similarity 0.88)
- Learning by Aligning Videos in Time (CVPR 2021 · similarity 0.87)
- Representation Learning via Global Temporal Alignment and Cycle-Consistency (CVPR 2021 · similarity 0.87)
- Learning To Align Sequential Actions in the Wild (CVPR 2022 · similarity 0.87)
- SyncVIS: Synchronized Video Instance Segmentation (NeurIPS 2024 · similarity 0.85)
- TTVidT: Decoupling the Temporal Axis for Efficient Motion-Centric Video Pretraining (NeurIPS 2026 · 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.
