基于离群值引导协同适应的注视估计泛化
计算机视觉与模式识别
2021-08-02 v2
摘要
深度神经网络显著提升了基于表观的注视估计精度。然而,将训练好的模型泛化到新域(如未知环境或人物)时,其性能仍不尽如人意。本文提出一种即插即用注视适应框架(PnP-GA),该框架为由离群值引导进行协同学习的网络集成。由于我们提出的框架不需要目标域中的真值标签,现有注视估计网络可直接插入PnP-GA并将算法泛化到新域。我们在四项注视域适应任务上测试PnP-GA:ETH-to-MPII、ETH-to-EyeDiap、Gaze360-to-MPII和Gaze360-to-EyeDiap。实验结果表明,PnP-GA框架相较基线系统分别取得36.9%、31.6%、19.4%和11.8%的显著性能提升。所提框架在注视域适应任务上也优于最先进的域适应方法。
引用
@article{arxiv.2107.13780,
title = {Generalizing Gaze Estimation with Outlier-guided Collaborative Adaptation},
author = {Yunfei Liu and Ruicong Liu and Haofei Wang and Feng Lu},
journal= {arXiv preprint arXiv:2107.13780},
year = {2021}
}
备注
This paper is accepted by ICCV2021. Code has been released at https://github.com/DreamtaleCore/PnP-GA