Research map / Video / Video-language understanding
VideoMem: Constructing, Analyzing, Predicting Short-Term and Long-Term Video Memorability
Romain Cohendet, Claire-Helene Demarty, Ngoc Q. K. Duong, Martin Engilberge
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 |
|---|---|---|---|
| use video-specific fine-tuned semantic embedding model instead of image-memorability MemNet applied to frames | video memorability prediction · VideoMem short-term (test) | Spearman rank correlation: 0.385 -> 0.494 | Improved |
| use semantic embedding model instead of MemNet | video memorability prediction · VideoMem long-term (test) | Spearman rank correlation: 0.168 -> 0.256 | Improved |
| fine-tune the full captioning model instead of frozen IC features plus MLP | video memorability prediction · VideoMem short-term (test) | Spearman rank correlation: 0.442 -> 0.494 | Improved |
| fine-tune semantic embedding model instead of fine-tuned ResNet101 | video memorability prediction · VideoMem short-term (test) | Spearman rank correlation: 0.46 -> 0.494 | Improved |
| fine-tune semantic embedding model instead of fine-tuned ResNet101 | video memorability prediction · VideoMem long-term (test) | Spearman rank correlation: 0.218 -> 0.256 | Improved |
| evaluate on the 500 most annotated test videos instead of the full test set | semantic embedding model · VideoMem short-term | Spearman rank correlation: 0.494 -> 0.565 | Improved |
| predict long-term instead of short-term memorability | semantic embedding model · VideoMem | Spearman rank correlation: 0.494 -> 0.256 | Hurt |
| use IC-based MLP on frozen features for long-term, evaluated on 500 most annotated videos | IC-based model · VideoMem long-term | Spearman rank correlation: 0.201 -> 0.188 | Hurt |
| count short-term score with long-term score | human memorability scores · VideoMem 10,000 videos | Spearman correlation: 0.305 | Mixed |
Closest papers by meaning
- VisMem: Latent Vision Memory Unlocks Potential of Vision-Language Models (CVPR 2026 · similarity 0.86)
- Captured by Captions: On Memorization and its Mitigation in CLIP Models (ICLR 2025 · similarity 0.86)
- FreshMem: Brain-Inspired Frequency-Space Hybrid Memory for Streaming Video Understanding (NeurIPS 2026 · similarity 0.85)
- Déjà Vu Memorization in Vision–Language Models (NeurIPS 2024 · similarity 0.85)
- VideoLucy: Deep Memory Backtracking for Long Video Understanding (NeurIPS 2025 · similarity 0.85)
- Measuring Dejavu Memorization Efficiently (NeurIPS 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.
