Skip to content
/ sst Public

My implementation (PyTorch) for the paper SST: Single-Stream Temporal Action Proposals (http://vision.stanford.edu/pdf/buch2017cvpr.pdf).

Notifications You must be signed in to change notification settings

pratik18v/sst

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

9 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

SST: Single-Stream Temporal Action Proposals

Pytorch implementation for the paper SST: Single-Stream Temporal Action Proposals!

SST is an efficient model for generating temporal action proposals in untrimmed videos. Analogous to object proposals for images, temporal action proposals provide the temporal bounds in videos where potential actions of interest may lie.


Resources

Quick links: [cvpr paper] [poster] [supplementary] [code]

Please use the following bibtex to cite our work:

@inproceedings{sst_buch_cvpr17,
  author = {Shyamal Buch and Victor Escorcia and Chuanqi Shen and Bernard Ghanem and Juan Carlos Niebles},
  title = {{SST}: Single-Stream Temporal Action Proposals},
  year = {2017},
  booktitle = {CVPR}
  }

As part of this repo, we also include evaluation notebooks, SST proposals for THUMOS'14, and pre-trained model parameters. Please see the code/ and data/ folders for more.

Dependencies

We include a requirements.txt file that lists all the dependencies you need. Once you have created a virtual environment, simply run pip install -r requirements.txt from within the environment to install all the dependencies. Note that the original code was executed using Python 2.7.