Automatic Facial Expression Recognition

Automatic Facial Expression Recognition

Project Membership:

• Leader: Prof. Christian W. Omlin

Staff:

Students:

• GARLIPP BGC
• EATWELL JC
• SIBIYA SB

Facial expressions play a dominant part in non-verbal communication and are therefore fundamental for automated analysis and understanding of human behaviour. It also has many useful applications in human/computer interactions. The objective is to develop an automatic facial expression recognition system using the facial action coding system (FACS) framework. It provides a method for describing individual facial muscle movements (so-called action units). We will use standard facial expression databases are freely available. The project consists of the following tasks:

Training of a suitable classifiers (e.g. support vector machines, convolutional neural networks) using labelled data of standard databases without feature extraction. Baseline performance analysis of the classifier. Training of a suitable classifier with extracted features and performance comparison. Performance comparisons between support vector machines and convolutional neural networks. Keywords: Computer Vision Facial Expression Recognition Neural Networks Human/Computer Interaction

How does the classifiers’ performances compare with and without feature extraction? Can the same classifiers be used for the detection of the intensity of facial expressions and how well do they perform? How does the performance of a standard classifier (support vector machine) compare with that of more purpose-built classifier (convolutional neural networks)? What are possible real-world applications?

Required resources: Personal computer Facial expression databases

##👏 Contributing I would love your help! Contribute by forking the repo and opening pull requests. Please ensure that your code passes the existing tests and linting, and write tests to test your changes if applicable.

All pull requests should be submitted to the main branch.