Aff-Wild-models

References

If you use any of the models/weights, please cite the following papers:

  1. D. Kollias, et. al.: "Deep Affect Prediction in-the-wild: Aff-Wild Database and Challenge, Deep Architectures, and Beyond". International Journal of Computer Vision (2019).

@article{kollias2019deep, title={Deep affect prediction in-the-wild: Aff-wild database and challenge, deep architectures, and beyond}, author={Kollias, Dimitrios and Tzirakis, Panagiotis and Nicolaou, Mihalis A and Papaioannou, Athanasios and Zhao, Guoying and Schuller, Bj{"o}rn and Kotsia, Irene and Zafeiriou, Stefanos}, journal={International Journal of Computer Vision}, pages={1--23}, year={2019}, publisher={Springer} }

  1. S. Zafeiriou, et. al. "Aff-Wild: Valence and Arousal in-the-wild Challenge", CVPRW, 2017.

@inproceedings{zafeiriou2017aff, title={Aff-wild: Valence and arousal ‘in-the-wild’challenge}, author={Zafeiriou, Stefanos and Kollias, Dimitrios and Nicolaou, Mihalis A and Papaioannou, Athanasios and Zhao, Guoying and Kotsia, Irene}, booktitle={Computer Vision and Pattern Recognition Workshops (CVPRW), 2017 IEEE Conference on}, pages={1980--1987}, year={2017}, organization={IEEE} }

  1. D. Kollias, et. al. "Recognition of affect in the wild using deep neural networks", CVPRW, 2017.

@inproceedings{kollias2017recognition, title={Recognition of affect in the wild using deep neural networks}, author={Kollias, Dimitrios and Nicolaou, Mihalis A and Kotsia, Irene and Zhao, Guoying and Zafeiriou, Stefanos}, booktitle={Computer Vision and Pattern Recognition Workshops (CVPRW), 2017 IEEE Conference on}, pages={1972--1979}, year={2017}, organization={IEEE} }

Pre-trained models:

The models on Aff-Wild can be downloaded from here.

Description:

The above link contains 3 folders named: "affwildnet-vggface-gru" , "affwildnet-resnet-gru" and "vggface".

The "vggface" folder contains two subfolders with 2 different models: both models are CNN networks based on VGG-FACE (with 3 fully connected layers with: i) 4096, 2000, 2 and ii) 4096, 4096, 2 units, respectively).

The "affwildnet-vggface-gru" folder contains the AffWildNet architecture (with no landmarks) as described in the paper entitled: "Deep Affect Prediction in-the-wild: Aff-Wild Database and Challenge, Deep Architectures, and Beyond".

The "affwildnet-resnet-gru" folder contains the AffWildNet architecture (with no landmarks and no fully connected layer; a Resnet-50 followed by a GRU network) as described in the paper entitled: "Deep Affect Prediction in-the-wild: Aff-Wild Database and Challenge, Deep Architectures, and Beyond".

Inside each of those folders, one can find the architectures of the networks, implemented in the Tensorflow environment and a readme explaining how to build/use them. An evaluation file is also uploaded with detailed explanation inside.

The Aff-Wild database can be downloaded from here. Specific details about the database and related stuff can be read in the Challenge's site.

Prerequisites:

  • The code works with Tensorflow 1.8
  • slim is also needed (it is incorporated within Tensorflow)