Skip to content

Github Code repo for the ICCV17 paper: 'Learning from Video and Text via Large-Scale Discriminative Clustering'

Notifications You must be signed in to change notification settings

antoine77340/iccv17learning

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

7 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Learning from Video and Text via Large-Scale Discriminative Clustering

Introduction

This is the code for the paper :

Antoine Miech, Jean-Baptiste Alayrac, Piotr Bojanowski, Ivan Laptev, Josef Sivic, Learning from Video and Text via Large-Scale Discriminative Clustering, ICCV17.

The webpage for this project is available here.

It only contains the code for the optimization part of the action recognition model given pre-extracted track features.

Contents

  1. Dependencies
  2. Data
  3. Demo
  4. Running on new images

Dependencies

To run this code, you need to install :

  1. MOSEK : version 7
  2. CVX : version 2.1

Once installed, setup the paths in the startup file :

main.m

Data

First you will need to download the pre-extracted person track features:

wget https://www.rocq.inria.fr/cluster-willow/amiech/iccv17/X.mat

Optimization

Now you can run our optimization code that will take X as input and output the label matrix Z given the bags formation and weak-supervision:

   main.m

This code is optimized for running everything on a computer with enough memory. If you are looking for a way to solve the Discriminative Clustering model in a totally online manner (ie with very limited memory usage) please contact me. We only provided this version as the fully online version is much slower to run because of the slow disk speed access.

Cite

If you find this code useful in your research, please, consider citing our paper:

@InProceedings{miech17learningvideotext, author = "Miech, Antoine and Alayrac, Jean-Baptiste and Bojanowski, Piotr and Laptev, Ivan and Sivic, Josef", title = "Learning from Video and Text via Large-Scale Discriminative Clustering", booktitle = "ICCV", year = "2017" }

About

Github Code repo for the ICCV17 paper: 'Learning from Video and Text via Large-Scale Discriminative Clustering'

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages