Research map / Video / Video segmentation and tracking
OpenSTL: A Comprehensive Benchmark of Spatio-Temporal Predictive Learning
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 |
|---|---|---|---|
| replace recurrent models with recurrent-free models | SimVP-style architecture · Moving MNIST | MSE: PredRNN++ 22.06 vs TAU 24.60 | Mixed |
| recurrent-free instead of recurrent (efficiency) | PredRNN++ vs MogaNet · Moving MNIST | FLOPs (G): 171.7 -> 16.5 | Helped |
| recurrent-free instead of recurrent (efficiency) | PredRNN++ vs VAN · Kitti&Caltech | FLOPs (G): 1803.0 -> 92.5 | Helped |
| recurrent-free VAN instead of recurrent PredRNN++ | spatio-temporal predictor · Kitti&Caltech | MSE: 125.5 vs 127.5 | Mixed |
| recurrent-free instead of recurrent models under missing frames | TAU vs PredRNN · Moving MNIST Missing | MSE: 46.05 -> 26.77 | Helped |
| recurrent-free instead of recurrent models under dynamic noise | TAU vs PredRNN++ · Moving MNIST Dynamic | MSE: 40.85 -> 43.37 | Hurt |
| recurrent-free instead of recurrent for weather forecasting | recurrent-free models · WeatherBench | MAE/RMSE | Helped |
| MetaFormer temporal module ViT instead of TAU | MetaVP · Moving MNIST | MSE: 24.60 -> 35.15 | Hurt |
| MetaFormer temporal module MogaNet instead of SimVP IncepU | MetaVP · Moving MNIST | MSE: 32.15 -> 25.57 | Helped |
Closest papers by meaning
- Wavelet-Driven Spatiotemporal Predictive Learning: Bridging Frequency and Time Variations (AAAI 2024 · similarity 0.88)
- Causal Spatio-Temporal Prediction: An Effective and Efficient Multi-Modal Approach (NeurIPS 2025 · similarity 0.87)
- UniSTD: Towards Unified Spatio-Temporal Learning across Diverse Disciplines (CVPR 2025 · similarity 0.86)
- How Different from the Past? Spatio-Temporal Time Series Forecasting with Self-Supervised Deviation Learning (NeurIPS 2025 · similarity 0.86)
- FSTLLM: Spatio-Temporal LLM for Few Shot Time Series Forecasting (ICML 2025 · similarity 0.86)
- Diffusion Transformers as Open-World Spatiotemporal Foundation Models (NeurIPS 2025 · 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.