Research map / Video / Action recognition
Anticipating Visual Representations From Unlabeled Video
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 three predicted futures instead of one (ablation) | Deep representation predictor · Activities of the Daily Living, five seconds before object appears | mean average precision: 9.1 -> 10.7 | Helped |
| Use three predicted futures instead of one (ablation) | Adapted Deep representation predictor · Activities of the Daily Living, five seconds before object appears | mean average precision: 8.7 -> 10.1 | Helped |
| Use three predicted futures instead of one (ablation) | Deep representation predictor · TV Human Interactions, one second before action | accuracy: 36.1 ± 6.4 -> 35.4 ± 5.2; 40.0 ± 4.9 -> 43.3 ± 4.7 | Mixed |
| Train the classifier on predicted representations (ablation) | Deep K=3 action predictor · TV Human Interactions, one second before action | accuracy: 35.4 ± 5.2 -> 43.3 ± 4.7 | Helped |
| Train the classifier on predicted representations (ablation) | Deep K=3 object predictor · Activities of the Daily Living, five seconds before object appears | mean average precision: 10.7 -> 10.1 | No clear effect |
| Predict ActionBank instead of fc7 (ablation) | Deep action predictor · TV Human Interactions, one second before action | accuracy: 35.4 ± 5.2 -> 35.7 ± 6.2; 43.3 ± 4.7 -> 35.7 ± 6.2 | Hurt |
| Train the predictor on THUMOS instead of the television-video repository (ablation) | Deep K=3 action predictor · TV Human Interactions, one second before action | accuracy: 35.4 ± 5.2 -> 29.1 ± 3.9; 43.3 ± 4.7 -> 43.6 ± 4.8 | Mixed |
| Adapt the classifier to the predicted representations (ablation) | Deep K=3 predictor trained on THUMOS · TV Human Interactions, one second before action | accuracy: 29.1 ± 3.9 -> 43.6 ± 4.8 | Helped |
Closest papers by meaning
- Deep Predictive Coding Networks for Video Prediction and Unsupervised Learning (ICLR 2017 · similarity 0.90)
- When Will You Do What? - Anticipating Temporal Occurrences of Activities (CVPR 2018 · similarity 0.90)
- Learning the Predictability of the Future (CVPR 2021 · similarity 0.90)
- Anticipating Future Relations via Graph Growing for Action Prediction (AAAI 2021 · similarity 0.88)
- ActFusion: a Unified Diffusion Model for Action Segmentation and Anticipation (NeurIPS 2024 · similarity 0.87)
- PlaySlot: Learning Inverse Latent Dynamics for Controllable Object-Centric Video Prediction and Planning (ICML 2025 · 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.