Research map / Video / Video generation
Learning To Cut by Watching Movies
Alejandro Pardo, Fabian Caba, Juan Léon Alcázar, Ali K. Thabet, Bernard Ghanem
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 auxiliary good-place-to-cut task (ablation) | Learning to Cut · Validation, d=1 | R@1K: 4.91 -> 8.18 | Improved |
| add auxiliary good-place-to-cut task (ablation) | Learning to Cut · Validation, d=1 | R@10K: 23.23 -> 30.40 | Improved |
| add two-stage inference (ATM top 30% then CLM) (ablation) | Learning to Cut single-stage · Validation, d=1 | R@1K: 2.89 -> 8.18 | Improved |
| add two-stage inference (ATM top 30% then CLM) (ablation) | Learning to Cut single-stage · Validation, d=3 | R@10K: 40.21 -> 79.42 | Improved |
| add audio stream (ablation) | Learning to Cut · Validation, d=1 | R@1K: 6.30 -> 8.18 | Improved |
| add audio stream (ablation) | Learning to Cut · Validation, d=1 | R@10K: 31.88 -> 30.40 | Hurt |
| add visual stream (ablation) | Learning to Cut · Validation, d=1 | R@1K: 7.82 -> 8.18 | Improved |
| add visual stream (ablation) | Learning to Cut · Validation, d=3 | R@10K: 84.13 -> 79.87 | Hurt |
| add audio features to visual raw features | raw backbone features · Validation, d=1 | R@5K: 2.78 -> 6.37 | Improved |
| learn contrastive projection on raw audio-visual features (single-stage) | Audio-visual Raw Features · Validation, d=1 | R@10K: 11.73 -> 16.15 | Improved |
| use raw visual features instead of random scoring | Random · Validation, d=1 | R@10K: 6.56 -> 5.17 | Hurt |
| use Learning to Cut on unedited footage instead of random | Random · Unedited footage, d=1 | R@1K: 0.83 -> 0.83 | No clear effect |
Closest papers by meaning
- MovieCuts: A New Dataset and Benchmark for Cut Type Recognition (ECCV 2022 · similarity 0.88)
- Señorita-2M: A High-Quality Instruction-based Dataset for General Video Editing by Video Specialists (NeurIPS 2025 · similarity 0.85)
- VEU-Bench: Towards Comprehensive Understanding of Video Editing (CVPR 2025 · similarity 0.84)
- Learning Action and Reasoning-Centric Image Editing from Videos and Simulation (NeurIPS 2024 · similarity 0.84)
- FADE: Frequency-Aware Diffusion Model Factorization for Video Editing (CVPR 2025 · similarity 0.84)
- EditBoard: Towards a Comprehensive Evaluation Benchmark for Text-Based Video Editing Models (AAAI 2025 · similarity 0.84)
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.
