Research map / Video / Action recognition
Look for the Change: Learning Object States and State-Modifying Actions From Untrimmed Web Videos
Check 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 causal ordering and noise-adaptive video weighting to learn states and actions from uncurated videos | Look for the Change · ChangeIt | action precision: 0.59 -> 0.68 | Helped |
| Use causal ordering and noise-adaptive video weighting to learn states and actions from uncurated videos | Look for the Change · ChangeIt | state precision: 0.30 -> 0.35 | Helped |
| Use the proposed learning method instead of the prior approach | Look for the Change · Alayrac et al. dataset | action precision: 0.55 -> 0.58 | Helped |
| Use the proposed learning method instead of the prior approach | Look for the Change · Alayrac et al. dataset | state precision: 0.48 -> 0.49 | Helped |
| Increase the training dataset from the top 5% toward the full dataset (ablation) | Look for the Change · ChangeIt | action precision | Helped |
| Increase the training dataset from the top 5% toward the full dataset (ablation) | Look for the Change · ChangeIt | state precision | Helped |
| Replace fixed-distance action negatives with uniformly random negatives (ablation) | Look for the Change · ChangeIt | action precision: 68% -> 62% | Hurt |
| Remove noise-adaptive video weighting (ablation) | Look for the Change · ChangeIt | action precision | Hurt |
| Remove noise-adaptive video weighting (ablation) | Look for the Change · ChangeIt | state precision | Hurt |
| Replace the two-layer MLP with a linear classifier (ablation) | Look for the Change · ChangeIt | action precision | Hurt |
| Remove input data augmentation (ablation) | Look for the Change · ChangeIt | action precision | Hurt |
| Remove the final CLIP projection layer from CLIP ViT-B/16 features (ablation) | Look for the Change · ChangeIt | state precision: 0.33 -> 0.35 | Helped |
Closest papers by meaning
- Learning Object State Changes in Videos: An Open-World Perspective (CVPR 2024 · similarity 0.90)
- Oops! Predicting Unintentional Action in Video (CVPR 2020 · similarity 0.88)
- Action Recognition From Single Timestamp Supervision in Untrimmed Videos (CVPR 2019 · similarity 0.88)
- Learning Temporal Co-Attention Models for Unsupervised Video Action Localization (CVPR 2020 · similarity 0.87)
- Learning Transferable Self-Attentive Representations for Action Recognition in Untrimmed Videos with Weak Supervision (AAAI 2019 · similarity 0.87)
- Weakly-Supervised Temporal Action Localization by Inferring Salient Snippet-Feature (AAAI 2024 · similarity 0.86)
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.