# Temporal Segment Networks (TSN)

Code & Models for Temporal Segment Networks (TSN) in ECCV 2016

Repository: https://github.com/yjxiong/temporal-segment-networks
Canonical: https://ross.abutalabs.com/products/temporal-segment-networks-tsn
Language: Python
License: BSD-2-Clause
License Family: permissive
Topics: temporal-segment-networks, action-recognition, video-understanding
Last push: 2020-10-27T15:37:48+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 3702, "days_push": 2136, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1577, forks 464 (observed 2026-08-28T04:05:06.108385+00:00)

## What it is
Official code and pretrained models for Temporal Segment Networks (TSN), a deep learning framework for video action recognition published at ECCV 2016/TPAMI 2018. It provides Caffe-based training/testing pipelines plus an experimental PyTorch implementation, with the authors recommending MMAction for new work.

## Use cases
- reproduce TSN action recognition results from the ECCV 2016 paper
- classify human actions in videos with pretrained Kinetics models
- train a video action recognition model on a custom dataset
- extract frames and optical flow from videos for training
- fine-tune TSN models for transfer learning on new video datasets
- compare video action recognition baselines for research

## When to choose
- you need the original TSN implementation to reproduce paper results
- you are a Caffe user needing maintained TSN support
- you want pretrained Kinetics action recognition weights from the authors
- you are studying or extending classic video understanding methods

## When to avoid
- you are starting a new action recognition project - use MMAction2 instead
- you need a maintained, modern PyTorch video toolbox
- you require Windows or non-Linux platform support
- you want production-ready video classification without research setup effort

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, video-processing, computer-vision
- domain: machine-learning, computer-vision, deep-learning
- platform: python
- tags: action-recognition, video-understanding, temporal-segment-networks, caffe, pytorch, research-code, optical-flow, eccv-2016, video, linux, docker, gpu

## Member repositories
- yjxiong/temporal-segment-networks (main) score 32
- yjxiong/tsn-pytorch (mirror) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:06.108385+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 2026-08-30T03:57:00.231711+00:00, confidence not recorded.
  - readme: https://github.com/yjxiong/temporal-segment-networks (fetched 2026-08-28T04:05:06.108385+00:00, sha e48655fda55f)
- Data as of 2026-08-30T08:39:29.467469+00:00.