Leveraging Large Face Recognition Data for Emotion Classification

Knyazev, Boris, Shvetsov, Roman, Efremova, Natalia and Kuharenko, Artem (2018) Leveraging Large Face Recognition Data for Emotion Classification. In: 2018 13th IEEE International Conference on Automatic Face & Gesture Recognition (FG 2018). IEEE, pp. 692-696. ISBN 978-1-5386-2335-0

Abstract

In this paper we describe a solution to our entry for the emotion recognition challenge EmotiW 2017. We propose an ensemble of several models, which capture spatial and audio features from videos. Spatial features are captured by convolutional neural networks, pretrained on large face recognition datasets. We show that usage of strong industry-level face recognition networks increases the accuracy of emotion recognition. Using our ensemble we improve on the previous year's best result on the test set by about 1%, achieving a 60.03% classification accuracy without any use of visual temporal information, showing a top-2 result in this challenge.

Item Type: Book Section
Subject(s): Technology
Centre: Oxford University Centre for Corporate Reputation
Date Deposited: 19 Jun 2018 09:27
Last Modified: 17 Oct 2018 14:13
Funders: not applicable
URI: http://eureka.sbs.ox.ac.uk/id/eprint/6852

Actions (login required)

Edit View Edit View