Improving Domain Generalization on Gaze Estimation via Branch-out Auxiliary Regularization

arXiv:2405.01439v1 Announce Type: new
Abstract: Despite remarkable advancements, mainstream gaze estimation techniques, particularly appearance-based methods, often suffer from performance degradation in uncontrolled environments due to variations in illumination and individual facial attributes. Existing domain adaptation strategies, limited by their need for target domain samples, may fall short in real-world applications. This letter introduces Branch-out Auxiliary Regularization (BAR), an innovative method designed to boost gaze estimation's generalization capabilities without requiring direct access to target domain data. Specifically, BAR integrates two auxiliary …

abstract applications arxiv cs.cv domain domain adaptation environments improving performance regularization samples strategies type via world