Research map / Video / Video segmentation and tracking
Stochastic Variational Video Prediction
Read the paperCheck 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 |
|---|---|---|---|
| Add stochastic latent variables to the deterministic video prediction model (ablation) | SV2P · BAIR robot pushing, action-free and action-conditioned | PSNR | Helped |
| Add stochastic latent variables to the deterministic video prediction model (ablation) | SV2P · Real-world video datasets | SSIM | Helped |
| Use time-variant rather than time-invariant latent sampling (ablation) | SV2P · BAIR robot pushing, extrapolation beyond the training horizon | PSNR | Mixed |
| Use time-invariant rather than time-variant latent sampling (ablation) | SV2P · Human3.6M | PSNR | Helped |
| Use time-invariant rather than time-variant latent sampling (ablation) | SV2P · Human3.6M | SSIM | Helped |
| Condition the inference network on future frames instead of using a filtering-based posterior conditioned only on context (ablation) | SV2P · Video prediction | latent utilization | Helped |
| Use the three-phase training procedure instead of naïve end-to-end training (ablation) | SV2P · BAIR robot pushing | reconstruction loss | Helped |
| Use naïve end-to-end training instead of the three-phase training procedure (ablation) | SV2P · BAIR robot pushing | reconstruction loss | Hurt |
| Sample latent values from the assumed prior rather than the approximated posterior (ablation) | SV2P · Stochastic movement dataset | motion direction accuracy | Hurt |
| Use SV2P instead of video pixel networks | Video prediction · Robotic pushing prediction | PSNR | Helped |
| Use SV2P instead of video pixel networks | Video prediction · Robotic pushing prediction | prediction noise | Helped |
| Use SV2P instead of the deterministic CDNA baseline | Video prediction · Human3.6M | human detection confidence | Helped |
Closest papers by meaning
- STDiff: Spatio-Temporal Diffusion for Continuous Stochastic Video Prediction (AAAI 2024 · similarity 0.87)
- Visual Representation Learning with Stochastic Frame Prediction (ICML 2024 · similarity 0.87)
- SyncVP: Joint Diffusion for Synchronous Multi-Modal Video Prediction (CVPR 2025 · similarity 0.86)
- Harmonizing Stochasticity and Determinism: Scene-responsive Diverse Human Motion Prediction (NeurIPS 2024 · similarity 0.85)
- Video Frame Prediction from a Single Image and Events (AAAI 2024 · similarity 0.85)
- Variational Inference for SDEs Driven by Fractional Noise (ICLR 2024 · similarity 0.84)
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.