LiteFer: An Approach Based on MobileViT Expression Recognition
Contact us if these videos are not relevant.
Contact us if these videos are not relevant.
View abstract on PubMed
Summary
This summary is machine-generated.LiteFer, a new lightweight network, efficiently recognizes facial expressions on mobile devices. It uses depth-separable convolution and attention to reduce size without losing accuracy, outperforming other methods.
Area Of Science
- Computer Vision
- Artificial Intelligence
- Machine Learning
Background
- Facial expression recognition is crucial for human-computer interaction.
- Complex Convolutional Neural Networks (CNNs) hinder deployment on resource-limited devices.
- Lightweight networks aim to reduce model size and parameters while maintaining accuracy.
Purpose Of The Study
- To develop a lightweight facial expression recognition method for mobile devices.
- To reduce network complexity and parameters without sacrificing recognition accuracy.
- To introduce the LiteFer method incorporating depth-separable convolution and attention.
Main Methods
- Implemented depth-separable convolution for efficient feature extraction.
- Integrated a lightweight attention mechanism to focus on salient facial features.
- Conducted comparative experiments on benchmark datasets (RAFDB, FERPlus).
Main Results
- LiteFer significantly reduces network parameters compared to existing methods.
- The proposed method demonstrates superior performance in facial expression recognition.
- Achieved high accuracy on both RAFDB and FERPlus datasets.
Conclusions
- LiteFer offers an effective solution for deploying facial expression recognition on edge devices.
- The method balances model efficiency with high recognition accuracy.
- LiteFer represents a significant advancement in lightweight deep learning for computer vision.
Contact us if these videos are not relevant.
Contact us if these videos are not relevant.