摘 要：随着人们日常生活娱乐活动多样性的增加，以及科技技术发展的快速步伐，虚拟角色的应用日益增多，通过深度学习技术的加持来助力虚拟角色产业的发展意义重大。文章对“深度学习”应用领域中的“虚拟角色”进行了研究，叙述了虚拟角色的制作过程，利用深度摄像头对虚拟角色进行肢体动作控制，基于深度学习技术生成虚拟角色脸部表情。
关键词：人工智能；虚拟角色；虚拟主播；动作捕捉；人脸表情生成
DOI:10.19850/j.cnki.2096-4706.2022.16.026
中图分类号：TP391.9 文献标识码：A 文章编号：2096-4706（2022）16-0100-04
Design and Implementation of Virtual Character Production System Based on Deep Learning
WU Zilong, XU Bingquan, DENG Xuanye
(Software Engineering Institute of Guangzhou, Guangzhou 510900, China)
Abstract: With the increase of diversity of people's recreations in daily life, and the rapid pace of the development of technology, the application of virtual characters is growing gradually. It is meaningful to make use of deep learning techniques to promote the development of virtual character industry. This paper studies the “virtual character” in the application field of “deep learning”, and describes the production process of the virtual character, uses the deep camera to control the limb movements of virtual characters, generates facial expressions of virtual characters based on deep learning technology.
Keywords: artificial intelligence; virtual character; virtual newscaster; motion capture; facial expression generation
参考文献：
[1] CUDEIRO D，BOLKART T，LAIDLAW C.Capture， Learning，and Synthesis of 3D Speaking Styles [J/OL].arXiv： 1905.03079 [cs.CV].[2022-04-25].https：//arxiv.org/abs/1905.03079.
[2] 郝昌 . 基于 AI+ 动作捕捉技术的虚拟主播体感交互系统的设计与实现 [J]. 广播与电视技术，2019，46（10)：48-52.
[3] SHOTTON J，FITZGIBBON A，COOK M. Real-time human pose recognition in parts from single depth images [J].Communications of the ACM，2013，56（1) ：116-124.
[4] ABADI M，AGARWAL A，BARHAM P，e t a l . TensorFlow：Large-Scale Machine Learning on Heterogeneous Distributed Systems [J/OL].arXiv：1603.04467 [cs.DC].（2016-05-16).https：//arxiv.org/abs/1603.04467.
[5] KINGMA D，BA J. Adam：A Method for Stochastic Optimization [J/OL].arXiv：1412.6980 [cs.LG].（2015-07-23).https：// arxiv.org/abs/1412.6980v8.
[6] Mozilla. Project DeepSpeech [EB/OL].[2022-04-27].https：// github.com/mozilla/DeepSpeech.
作者简介：吴梓龙（2000.11—），男，汉族，广东潮州人，本科在读，研究方向：虚拟现实技术开发；徐炳权（2001.05—），男，汉族，广东汕尾人，学士学位，研究方向：虚拟现实技术开发；邓炫烨（2002.01—），女，汉族，广东中山人，学士学位，研究方向：虚拟现实技术开发。