Research map / Video / Video segmentation and tracking
LAMV: Learning to Align and Match Videos With Kernelized Temporal Layers
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 |
|---|---|---|---|
| Use LAMV with both vector and frequency normalization instead of CTE | Video alignment with RMAC · Madonna | queries localized better than 1s (%): 51.4 -> 76.1 | Helped |
| Use LAMV with both vector and frequency normalization instead of CTE | Video alignment with RMAC · Climbing | queries localized better than 1s (%): 24.7 -> 34.8 | Helped |
| Use LAMV with both normalizations instead of TMK | Video copy detection · VCDB | maximum segment-level F1 score: 67.4 -> 68.7 | Helped |
| Use LAMV instead of TMK | Event retrieval · EVVE | mAP: 51.6 -> 53.6 | Helped |
| Add average query expansion | LAMV with RMAC · EVVE | mAP: 53.6 -> 58.7 | Helped |
| Use both vector and frequency normalization instead of frequency normalization only (ablation) | LAMV with RMAC · EVVE | mAP: 53.5 -> 53.6 | No clear effect |
| Add frequency normalization to vector normalization (ablation) | LAMV with RMAC · EVVE | mAP: 51.9 -> 53.6 | Helped |
| Combine multiple periods rather than use only a long period (ablation) | LAMV video alignment · Madonna | localization accuracy | Helped |
| Use only short periods rather than a mix including medium-long periods (ablation) | LAMV video alignment · Madonna | retrieval performance | Hurt |
| Train the full architecture end to end (ablation) | LAMV · Video matching and alignment | performance | No clear effect |
Closest papers by meaning
- DejaVid: Encoder-Agnostic Learned Temporal Matching for Video Classification (CVPR 2025 · similarity 0.87)
- Temporal Alignment-Free Video Matching for Few-shot Action Recognition (CVPR 2025 · similarity 0.86)
- Anchor-Aware Similarity Cohesion in Target Frames Enables Predicting Temporal Moment Boundaries in 2D (CVPR 2025 · similarity 0.86)
- VidLA: Video-Language Alignment at Scale (CVPR 2024 · similarity 0.85)
- From Play to Replay: Composed Video Retrieval for Temporally Fine-Grained Videos (NeurIPS 2025 · similarity 0.85)
- Align Before Adapt: Leveraging Entity-to-Region Alignments for Generalizable Video Action Recognition (CVPR 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.