Emotion_Recognition_using_Facial_Landmarks (1).pdf (537 kB)
Download file

An Adversarial Attacks Resistance-based Approach to Emotion Recognition from Images using Facial Landmarks

Download (537 kB)
conference contribution
posted on 14.04.2021, 01:35 by Harisu Abdullahi ShehuHarisu Abdullahi Shehu, William Browne, Hedwig EisenbarthHedwig Eisenbarth
Emotion recognition has become an increasingly important area of research due to the increasing number of CCTV cameras in the past few years. Deep network-based methods have made impressive progress in performing emotion recognition-based tasks, achieving high performance on many datasets and their related competitions such as the ImageNet challenge. However, deep networks are vulnerable to adversarial attacks. Due to their homogeneous representation of knowledge across all images, a small change to the input image made by an adversary might result in a large decrease in the accuracy of the algorithm. By detecting heterogeneous facial landmarks using the machine learning library Dlib we hypothesize we can build robustness to adversarial attacks. The residual neural network (ResNet) model has been used as an example of a deep learning model. While the accuracy achieved by ResNet showed a decrease of up to 22%, our proposed approach has shown strong resistance to an attack and showed only a little (< 0.3%) or no decrease when the attack is launched on the data. Furthermore, the proposed approach has shown considerably less execution time compared to the ResNet model.

History

Preferred citation

Shehu, H. A., Browne, W. & Eisenbarth, H. (2020, August). An Adversarial Attacks Resistance-based Approach to Emotion Recognition from Images using Facial Landmarks. In 29th IEEE International Conference on Robot and Human Interactive Communication, RO-MAN 2020 2020 29th IEEE International Conference on Robot and Human Interactive Communication (RO-MAN) (00 pp. 1307-1314). IEEE. https://doi.org/10.1109/RO-MAN47096.2020.9223510

Conference name

2020 29th IEEE International Conference on Robot and Human Interactive Communication (RO-MAN)

Conference start date

31/08/2020

Conference finish date

04/09/2020

Title of proceedings

29th IEEE International Conference on Robot and Human Interactive Communication, RO-MAN 2020

Volume

00

Publication or Presentation Year

01/08/2020

Pagination

1307-1314

Publisher

IEEE

Publication status

Published

ISSN

1944-9445

Usage metrics

Categories

Exports