EM-Net: Gaze Estimation with Expectation Maximization Algorithm

TL;DR

EM-Net is a lightweight gaze estimation model that combines deep learning with Expectation Maximization, improving accuracy and robustness while reducing data requirements and computational resources.

Contribution

The paper introduces EM-Net, a novel gaze estimation model integrating a global attention mechanism and EM algorithm for enhanced performance and generalization with less data.

Findings

EM-Net improves accuracy on multiple datasets by over 2%.

The model maintains robustness under Gaussian noise.

Uses only 50% of training data for comparable performance.

Abstract

In recent years, the accuracy of gaze estimation techniques has gradually improved, but existing methods often rely on large datasets or large models to improve performance, which leads to high demands on computational resources. In terms of this issue, this paper proposes a lightweight gaze estimation model EM-Net based on deep learning and traditional machine learning algorithms Expectation Maximization algorithm. First, the proposed Global Attention Mechanism(GAM) is added to extract features related to gaze estimation to improve the model's ability to capture global dependencies and thus improve its performance. Second, by learning hierarchical feature representations through the EM module, the model has strong generalization ability, which reduces the need for sample size. Experiments have confirmed that, on the premise of using only 50% of the training data, EM-Net improves the performance of Gaze360, MPIIFaceGaze, and RT-Gene datasets by 2.2%, 2.02%, and 2.03%, respectively, compared with GazeNAS-ETH. It also shows good robustness in the face of Gaussian noise interference.