Facial Expression Recognition from Multi-Perspective Visual Inputs and Soft Voting.

Area of Science:

• Computer Vision
• Machine Learning
• Human-Computer Interaction

Background:

• Facial expression recognition is crucial for applications like mental health monitoring and user feedback.
• Current methods often rely on Machine Learning (ML) techniques for data-driven development.
• Previous research utilized human-invented rules, which are less adaptable than ML.

Purpose of the Study:

• To develop an advanced Machine Learning-based method for automatic facial expression recognition.
• To leverage information fusion architecture and soft voting for improved performance.
• To demonstrate the superiority of fusion architectures over standard feature aggregation in ML.

Main Methods:

• Utilized visual information from a Kinect device, specifically a dataset of face points.
• Implemented a Machine Learning approach incorporating information fusion architecture techniques.
• Applied soft voting as a mechanism to combine multiple information sources.

Main Results:

• Achieved average prediction performance significantly exceeding state-of-the-art results on the considered dataset.
• Demonstrated the effectiveness of the proposed information fusion architecture.
• Confirmed that fusion architectures are more advantageous than basic feature aggregation in ML.

Conclusions:

• The proposed Machine Learning method with information fusion is highly effective for facial expression recognition.
• Information fusion architectures offer a significant advantage over traditional feature aggregation methods.
• This approach advances the field of automatic facial expression recognition with practical implications.