当前位置: X-MOL 学术Interact. Comput. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
EM-FEE: An Efficient Multitask Scheme for Facial Expression Estimation
Interacting with Computers ( IF 1.3 ) Pub Date : 2020-06-03 , DOI: 10.1093/iwcomp/iwaa011
Bin Yang 1 , Zhenyu Li 1 , Yingtao Sun 1 , Enguo Cao 1
Affiliation  

Human–computer interaction (HCI) has received growing interest in both academic research and the design of information technological applications. Automated facial expression estimation of image is a difficult, yet crucial, problem in the design of HCI system. Although artificial neural network has achieved many remarkable results, few smart wearable devices can benefit from it. Most of these devices are constrained by limited computing and storage capacity. An effective solution is to allow servers to handle multiple tasks simultaneously. Toward this goal, we have been building an Efficient multitask scheme for facial expression estimation (EM-FEE). A multitask neural network is designed to enable the HCI system to accomplish different related tasks at the same time, that is, locating the user’s facial landmarks and estimating facial expressions. Experimental results demonstrate that our proposed scheme outperforms state-of-the-art. Finally, we review the remaining challenges and corresponding opportunities as well as future directions of the design of facial expression estimation systems for smart wearable devices.

中文翻译:

EM-FEE:高效的多任务面部表情估计方案

人机交互(HCI)在学术研究和信息技术应用程序设计中越来越受到关注。在HCI系统的设计中,图像的自动面部表情估计是一个困难但至关重要的问题。尽管人工神经网络已经取得了许多显着的成果,但是很少有智能穿戴设备可以从中受益。这些设备中的大多数都受到有限的计算和存储容量的限制。一个有效的解决方案是允许服务器同时处理多个任务。为了实现这一目标,我们一直在建立一种高效的多任务面部表情估计方案(EM-FEE)。多任务神经网络被设计为使HCI系统能够同时完成不同的相关任务,即定位用户的面部标志和估计面部表情。实验结果表明,我们提出的方案优于最新技术。最后,我们回顾了智能可穿戴设备的面部表情估计系统设计所面临的挑战和机遇,以及未来的发展方向。
更新日期:2020-06-03
down
wechat
bug