# chengtan9907/OpenSTL

OpenSTL: A Comprehensive Benchmark of Spatio-Temporal Predictive Learning

Repository: https://github.com/chengtan9907/OpenSTL
Canonical: https://ross.abutalabs.com/products/openstl
Homepage: https://openstl.readthedocs.io/en/latest/
Language: Python
License: Apache-2.0
License Family: permissive
Topics: benchmark, deep-learning, predictive-learning, pytorch, self-supervised-learning, video-prediction, weather-forecast, awesome-list, artificial-intelligence, attention-mechanism, computer-vision, mlp, transformer, awesome-lists
Last push: 2026-03-01T17:25:49+00:00

## Health v2 (maintenance only)
Score: 54/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 70, release rhythm 8, longevity 100
- inputs: {"age_days": 1498, "days_push": 185, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1137, forks 191 (observed 2026-08-28T04:03:43.935990+00:00)

## What it is
OpenSTL is a comprehensive benchmark and modular framework for spatio-temporal predictive learning, covering video prediction methods across synthetic and real-world tasks like human motion, driving scenes, traffic flow, and weather forecasting. It provides a layered PyTorch codebase with a PyTorch Lightning implementation, model zoo, and visualization tools.

## Use cases
- benchmark video prediction models
- predict future weather frames from radar data
- forecast traffic flow from video
- train spatio-temporal predictive models in PyTorch
- compare transformer and CNN video prediction methods
- run experiments on moving MNIST and KTH datasets

## When to choose
- you need a standardized benchmark for spatio-temporal prediction research
- you want a modular framework to implement and compare video prediction algorithms
- you need pretrained model zoos for weather or traffic forecasting

## When to avoid
- you need production video analytics rather than research benchmarking
- your task is single-image prediction without temporal dynamics
- you work outside PyTorch

## Facets
- artifact type: framework
- maturity: active
- function: machine-learning, deep-learning, benchmarking, data-science
- domain: deep-learning, computer-vision, artificial-intelligence, data-science
- platform: python
- tags: video-prediction, spatio-temporal, weather-forecasting, pytorch-lightning, self-supervised-learning, model-zoo, benchmark-suite, gpu

## Member repositories
- chengtan9907/OpenSTL (main) score 54

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:43.935990+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T06:36:14.211114+00:00, confidence not recorded.
  - readme: https://github.com/chengtan9907/OpenSTL (fetched 2026-08-28T04:03:43.935990+00:00, sha e7fd0c9f9034)
- Data as of 2026-08-30T08:39:29.467469+00:00.