Description/Abstract/Artist Statement

Studying facial expressions can provide insight into the development of social skills in children and provide support to individuals with developmental disorders. In afflicted individuals, such as children with Autism Spectrum Disorder (ASD), atypical interpretations of facial expressions are well-documented. In computer vision, many popular and state-of-the-art deep learning architectures (VGG16, EfficientNet, ResNet, etc.) are readily available with pre-trained weights for general object recognition. Transfer learning utilizes these pre-trained models to improve generalization on a new task. In this project, transfer learning is implemented to leverage the pretrained model (general object recognition) on facial expression classification. Though this method, the base and middle layers are preserved to exploit the existing neural architecture. The investigated method begins with a base-packaged architecture trained on ImageNet. This foundation is then task changed from general object classification to facial expression classification in the first transfer learning step. The second transfer learning step performs a domain change from adult to child data. Finally, the trained network is evaluated on the child facial expression classification task.

Presenting Author Name/s

Gregory Hubbard

Faculty Advisor/Mentor

Khan Iftekharuddin

College Affiliation

College of Engineering & Technology (Batten)

Presentation Type

Poster

Disciplines

Artificial Intelligence and Robotics | Bioimaging and Biomedical Optics | Biomedical Engineering and Bioengineering

Session Title

Poster Session

Location

Learning Commons @ Perry Library

Start Date

3-19-2022 9:00 AM

End Date

3-19-2022 11:00 AM

Upload File

wf_yes

Share

COinS
 
Mar 19th, 9:00 AM Mar 19th, 11:00 AM

Two-Stage Transfer Learning for Facial Expression Classification in Children

Learning Commons @ Perry Library

Studying facial expressions can provide insight into the development of social skills in children and provide support to individuals with developmental disorders. In afflicted individuals, such as children with Autism Spectrum Disorder (ASD), atypical interpretations of facial expressions are well-documented. In computer vision, many popular and state-of-the-art deep learning architectures (VGG16, EfficientNet, ResNet, etc.) are readily available with pre-trained weights for general object recognition. Transfer learning utilizes these pre-trained models to improve generalization on a new task. In this project, transfer learning is implemented to leverage the pretrained model (general object recognition) on facial expression classification. Though this method, the base and middle layers are preserved to exploit the existing neural architecture. The investigated method begins with a base-packaged architecture trained on ImageNet. This foundation is then task changed from general object classification to facial expression classification in the first transfer learning step. The second transfer learning step performs a domain change from adult to child data. Finally, the trained network is evaluated on the child facial expression classification task.