Research map / Video / Video-language understanding
AutoTransition: Learning to Recommend Video Transition Effects
Yaojie Shen, Libo Zhang, Kai Xu, Xiaojie Jin
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 sequential context (ablation) | AutoTransition visual · transition dataset | Recall@1: 24.12% -> 25.40% | Improved |
| add audio input to visual+context (ablation) | AutoTransition · transition dataset | Recall@1: 25.40% -> 28.06% | Improved |
| use visual input instead of audio as input (ablation) | AutoTransition · transition dataset | Recall@1: 19.39% -> 25.40% | Improved |
| use pre-trained transition embedding with projection instead of random embedding (ablation) | AutoTransition · transition dataset | Recall@1: 25.67% -> 28.06% | Improved |
| add linear projection to pre-trained embedding (ablation) | AutoTransition · transition dataset | Recall@1: 26.24% -> 28.06% | Improved |
| use pre-trained embedding without projection instead of random (ablation) | AutoTransition · transition dataset | Recall@5: 66.3% -> 66.03% | No clear effect |
| use embedding matching with triplet loss instead of classification (ablation) | AutoTransition · transition dataset | Recall@1: 22.27% -> 28.06% | Improved |
| unfreeze SlowFast stages 1-3 versus freezing them (ablation) | AutoTransition visual-only · transition dataset | Recall@1: 25.40% -> 25.97% | Improved |
| freeze stage 1-3 instead of all parameters (ablation) | AutoTransition visual-only · transition dataset | Recall@5: 26.53% -> 66.33% | Improved |
| reduce matching dimension from 2048 to 512 (ablation) | AutoTransition · transition dataset | Recall@1: 28.06% -> 26.40% | Hurt |
| increase transformer layers from 2 to 4 (ablation) | AutoTransition · transition dataset | Recall@1: 28.06% -> 27.26% | Hurt |
| increase d_model from 2048 to 4096 (ablation) | AutoTransition · transition dataset | Recall@1: 28.06% -> 26.93% | Hurt |
Later papers that compared with this method
1 later accepted paper reports AutoTransition as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- V-Trans4Style: Visual Transition Recommendation for Video Production Style Adaptation (ECCV 2024 · better on 2 of 3 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- SEINE: Short-to-Long Video Diffusion Model for Generative Transition and Prediction (ICLR 2024 · similarity 0.86)
- Learning to Highlight Audio by Watching Movies (CVPR 2025 · similarity 0.85)
- COVE: Unleashing the Diffusion Feature Correspondence for Consistent Video Editing (NeurIPS 2024 · similarity 0.85)
- AVENUE: Audio-Video Editing Understanding and Evaluation (NeurIPS 2026 · similarity 0.84)
- VideoGuide: Improving Video Diffusion Models without Training Through a Teacher's Guide (CVPR 2025 · similarity 0.84)
- Zero-shot Image Editing with Reference Imitation (NeurIPS 2024 · 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.
