Skip to content

dkollias/OMG_Challenge

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

13 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

OMG_Challenge

This repository contains our solution to the OMG-Emotion Challenge 2018 -using only visual data- that ranked 2nd for vision-only valence estimation and 2nd for overall valence estimation.

If you use our scripts please cite the papers:

  1. A Multi-component CNN-RNN Approach for Dimensional Emotion Recognition in-the-wild

@article{kollias2018multi, title={A Multi-component CNN-RNN Approach for Dimensional Emotion Recognition in-the-wild}, author={Kollias, Dimitrios and Zafeiriou, Stefanos}, journal={arXiv preprint arXiv:1805.01452}, year={2018} }

  1. Exploiting multi-CNN features in CNN-RNN based Dimensional Emotion Recognition on the OMG in-the-wild Dataset

@article{kollias2019exploiting, title={Exploiting multi-cnn features in cnn-rnn based dimensional emotion recognition on the omg in-the-wild dataset}, author={Kollias, Dimitrios and Zafeiriou, Stefanos}, journal={arXiv preprint arXiv:1910.01417}, year={2019} }

This repository contains: train and evaluation scripts (train_script.py and eval_script.py), their models (AffWildNet.py and AffWildNet_valid.py) and a script used by the train and evaluation scripts for processing the data (data_process.py) . To be more exact each script has flags (one can find more detailed explanation withing each script) :

  • train_script flags:

             initial_learning_rate 
             
             concordance_loss : if 1 use concordance as loss function, else if 0: use MSE as loss function
             
             batch_size
             
             seq_length
             
             size : size of input images, e.g. if is set to 96 then input_images_size = 96x96
             
             h_units
             
             network : which network to use, pick between: "CNN_GRU_1RNN", "CNN_GRU_3RNN"  
             
             input_file
             
             train_dir
             
             pretrained_model_checkpoint_path 
    
  • eval_script flags:

             batch_size
             
             seq_length
             
             size : size of input images, e.g. if is set to 96 then input_images_size = 96x96
             
             h_units
             
             network : which network to use, pick between: "CNN_GRU_1RNN", "CNN_GRU_3RNN"  
             
             input_file
             
             pretrained_model_checkpoint_path 
    

Dependencies:

numpy : we are using version 1.13.1

tensorflow: we are using version 1.1.0

(we also use tensorflow.contrib.slim : slim is part of tensorflow after version 1.0)

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages