Research map / Video / Action recognition
Transferring Knowledge From Text to Video: Zero-Shot Anticipation for Procedural Actions
Fadime Şener, Rishabh Saraf, Angela Yao
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 |
|---|---|---|---|
| pre-train the text model on more of Recipe1M (25% to 100%) (ablation) | ours text · Tasty Videos V1 | BLEU4: 2.03 -> 3.30 | Improved |
| pre-train on 50% instead of 0% of Recipe1M (ablation) | ours text · Tasty Videos V1 | BLEU4: 0.24 -> 2.42 | Improved |
| add recipe loss L_r to decoder loss L_d (ablation) | recipe network · Recipe1M test, next step | ingredient recall: 32.3 -> 33.5 | Improved |
| add recipe loss L_r to decoder loss L_d (ablation) | recipe network · Recipe1M test, next+3 step | BLEU4: 1.3 -> 0.4 | Hurt |
| use beam search (k=5) instead of greedy decoding (ablation) | recipe network · Recipe1M test, next step | ingredient recall: 32.3 -> 33.1 | Improved |
| use ground-truth segments instead of fixed windows (ablation) | ours visual · Tasty Videos V1 | BLEU4: 1.23 -> 1.48 | Improved |
| remove the ingredient list input (ablation) | ours text · Tasty Videos V1 | BLEU4: 3.30 -> 0.92 | Hurt |
| add accompanying text with video (joint training) (ablation) | ours visual (window) · Tasty Videos V1 | BLEU4: 1.23 -> 2.33 | Improved |
| use segment proposals instead of fixed windows (ablation) | ours video · Tasty Videos V2 | BLEU4: 1.42 -> 1.63 | Improved |
| use Recipe1M pre-training in supervised setting (ablation) | sup. visual (GT) · YouCookII 4-fold | BLEU4: 0.15 -> 1.21 | Improved |
| use zero-shot instead of supervised training on same dishes (ablation) | visual (GT) · YouCookII 4-fold | BLEU4: 1.21 -> 0.84 | Hurt |
| anticipate next step versus caption current step with S2VT | S2VT · Tasty Videos V1 | BLEU4: 1.10 -> 0.26 | Hurt |
Closest papers by meaning
- Can't Make an Omelette Without Breaking Some Eggs: Plausible Action Anticipation Using Large Video-Language Models (CVPR 2024 · similarity 0.89)
- RoboDreamer: Learning Compositional World Models for Robot Imagination (ICML 2024 · similarity 0.88)
- Solving New Tasks by Adapting Internet Video Knowledge (ICLR 2025 · similarity 0.87)
- VideoWorld: Exploring Knowledge Learning from Unlabeled Videos (CVPR 2025 · similarity 0.87)
- Procedure Knowledge Decoupled Distillation Strategy for Procedure Planning in Instructional Videos (AAAI 2025 · similarity 0.87)
- Seer: Language Instructed Video Prediction with Latent Diffusion Models (ICLR 2024 · similarity 0.87)
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.
