Research map / Video / Video segmentation and tracking
Self-Supervised Spatio-Temporal Representation Learning for Videos by Predicting Motion and Appearance Statistics
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 |
|---|---|---|---|
| pre-train C3D with motion statistics instead of random initialization | C3D · UCF101 | accuracy (%): 45.4 -> 57.8 | Helped |
| use global motion statistics alongside all local motion patterns (ablation) | C3D motion-statistics pre-training · UCF101 | accuracy (%): 55.4 -> 57.8 | Helped |
| use global motion statistics only rather than random initialization (ablation) | C3D · UCF101 | accuracy (%): 45.4 -> 48.3 | Helped |
| use motion pattern 1 pre-training rather than random initialization (ablation) | C3D · UCF101 | accuracy (%): 45.4 -> 53.8 | Helped |
| use motion pattern 2 pre-training rather than random initialization (ablation) | C3D · UCF101 | accuracy (%): 45.4 -> 53.2 | Helped |
| use motion pattern 3 pre-training rather than random initialization (ablation) | C3D · UCF101 | accuracy (%): 45.4 -> 54.2 | Helped |
| combine motion and appearance statistics rather than motion statistics alone (ablation) | C3D self-supervised pre-training · UCF101 | accuracy (%): 57.8 -> 58.8 | Helped |
| combine motion and appearance statistics rather than motion statistics alone (ablation) | C3D self-supervised pre-training · HMDB51 | accuracy (%): 29.95 -> 32.6 | Helped |
| pre-train with appearance statistics only rather than random initialization (ablation) | C3D · UCF101 | accuracy (%): 45.4 -> 48.6 | Helped |
| pre-train with appearance statistics only rather than random initialization (ablation) | C3D · HMDB51 | accuracy (%): 19.7 -> 20.3 | Helped |
| pre-train with motion statistics only rather than appearance statistics only (ablation) | C3D · UCF101 | accuracy (%): 48.6 -> 57.8 | Helped |
| pre-train with motion statistics only rather than appearance statistics only (ablation) | C3D · HMDB51 | accuracy (%): 20.3 -> 29.95 | Helped |
Closest papers by meaning
- Spatiotemporal Contrastive Video Representation Learning (CVPR 2021 · similarity 0.90)
- Spatial-Then-Temporal Self-Supervised Learning for Video Correspondence (CVPR 2023 · similarity 0.90)
- No More Shortcuts: Realizing the Potential of Temporal Self-Supervision (AAAI 2024 · similarity 0.89)
- A Large-Scale Study on Unsupervised Spatiotemporal Representation Learning (CVPR 2021 · similarity 0.89)
- Video Playback Rate Perception for Self-Supervised Spatio-Temporal Representation Learning (CVPR 2020 · similarity 0.89)
- Contrastive Spatio-Temporal Pretext Learning for Self-Supervised Video Representation (AAAI 2022 · similarity 0.89)
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.