淅川旅游网

淅川旅游网

http://www.xclyw.com

在“深度换脸”的领域内,众人的注意力主要集中在可能对政治现实造成严重破坏的“虚假面孔”上,以及那些刻意
菜单导航

什么?!有人模仿你的脸,还有人模仿你全身?

作者: 大成 发布时间: 2021年01月19日 00:36:12

You’ve been warned: Full body deepfakes are the next step in AI-based human mimicry

数据观丨王婕(译)

什么?!有人模仿你的脸,还有人模仿你全身?

  This developing branch of synthetic media technology has commercial applications—but also has the potential to disrupt elections and spread disinformation.

  以下这项发展中的合成媒体技术分支具备商业用途,但也有可能被用来扰乱选举和散布谣言。

  In Russian novelist Victor Pelevin’s cyberpunk novel, Homo Zapiens, a poet named Babylen Tatarsky is recruited by an old college buddy to be an advertising copywriter in Moscow amid post-Soviet Russia’s economic collapse. With a talent for clever wordplay, Tatarsky quickly climbs the corporate ladder, where he discovers that politicians like then-Russian president Boris Yeltsin and major political events are, in fact, virtual simulations. With the advent of ever-more sophisticated deepfakes, it feels as if something like Pelevin’s vision is slowly coming true.

  俄罗斯小说家维克多·佩列文(Victor Pelevin)的赛博朋克小说《Homo Zapiens》中,一位名叫Babylen Tatarsky的诗人在苏联解体,俄罗斯经济面临崩溃之际,被一位在莫斯科的大学老友聘为广告文案撰稿人。Tatarsky凭借着巧妙的文字天赋一路水涨船高,而他也逐渐发现了这样一个事实:时任俄罗斯总统的叶利钦等政要和当时的重大政治事件,实际上都是虚拟仿真的产物。放眼现在,随着日益纯熟的“深度换脸”技术出现,佩列文的想象似乎正在慢慢变为现实。

  (数据观注释:【赛博朋克小说】赛博朋克小说属于科幻小说的类型,兴起于上世纪七十年代的美国,这一类故事里有大量对新兴信息技术和生物科技的描写,常常涉及跨国财团垄断高新技术,故事的主角一般会设定成游走在社会主流之外的边缘人,他们活在未来社会的阴暗面,喜欢修改电脑的软硬件配置,崇尚改造身体,拒绝融入主流体制,靠着合法或者非法的技术手段铤而走险,有时不惜与超级大公司对抗。这种高与低并存产生的反差,造成了一种特殊的美学效果,被概括为“高科技、低生活”六个字。)

  Within the field of deepfakes, or “synthetic media” as researchers call it, much of the attention has been focused on fake faces potentially wreaking havoc on political reality, as well as other deep learning algorithms that can, for instance, mimic a person’s writing style and voice. But yet another branch of synthetic media technology is fast evolving: full body deepfakes.

  在“深度换脸”(亦或被研究人员称之为“合成媒体”)的领域内,众人的注意力主要集中在可能对政治现实造成严重破坏的“虚假面孔”上,以及那些刻意模仿人写作风格和声音的深度学习算法。然而,如今合成媒体技术的另一个分支——“深度换身”正在迅速发展。

  In August 2018, University of California Berkeley researchers released a paper and video titled “Everybody Dance Now,” demonstrating how deep learning algorithms can transfer a professional dancers’ moves onto the bodies of amateurs. While primitive, it showed that machine learning researchers are tackling the more difficult task of creating full body deepfakes. Also in 2018, a team of researchers led by Dr. Björn Ommer of Heidelberg University in Germany published a paper on teaching machines to realistically render human movements. And in April of this year, the Japanese artificial intelligence company Data Grid developed an AI that can automatically generate whole body models of nonexistent persons, identifying practical applications in the fashion and apparel industries.

  2018年8月,加州大学伯克利分校的研究人员发表了一篇题为《人人都在跳舞》的论文和视频,展示了深度学习算法如何将专业舞者的动作转移到业余舞者身上。虽然这一研究成果还有待完善,但已表明机器学习的研究人员正在着手更具挑战的任务——“深度换身”。同年,德国海德堡大学的比约恩·奥默博士领导的一个研究团队发表了一篇关于教会机器逼真还原人类动作的论文。今年4月,日本人工智能公司Data Grid开发了一种人工智能技术,可以自动生成不存在的人体全身模型,并证实了它在时尚和服装领域中的实际应用。

  While it’s clear that full body deepfakes have interesting commercial applications, like deepfake dancing apps or in fields like athletics and biomedical research, malicious use cases are an increasing concern amid today’s polarized political climate riven by disinformation and fake news. For now, full body deepfakes aren’t capable of completely fooling the eye, but like any deep learning technology, advances will be made. It’s only a question of how soon full body deepfakes will become indistinguishable from the real.

声明:本媒体部分图片、文章来源于网络,版权归原作者所有,如有侵权,请与我联系删除。