ADM-201 dump PMP dumps pdf SSCP exam materials CBAP exam sample questions

新型透视镜:隔着墙也能看到你的新技术 – 译学馆
未登陆,请登陆后再发表信息
最新评论 (0)
播放视频

新型透视镜:隔着墙也能看到你的新技术

This AI Senses Humans Through Walls

各位学者 这是一篇时长两分钟的学术论文
Dear Fellow Scholars, this is Two Minute Paperswith Károly Zsolnai-Fehér.
姿势预测是个有趣的研究领域
Pose estimation is an interesting area
通常我们会收集一些关于人类的
of research where we typically have a few images
图像或短视频
or video footage of humans,
然后尝试自动提取一个人的姿势
and we try to automatically extract the pose a person was taking.
简而言之 输入一张或多张图片
In short, the input is one or more photo,
就可以输出一副人体骨架
and the output is typically a skeleton of the person.
那么 它用于何处呢?
So what is this good for?
可以用于很多方面
A lot of things.
例如 我们可以利用这些骨架
For instance, we can use these skeletons
以极低的成本将人类姿势转化至
to cheaply transfer the gestures of a human onto
虚拟人物的身上 还有老年人跌倒检测
a virtual character, fall detection for the elderly,
运动员的运动分析
analyzing the motion of athletes,
以及其他众多领域
and many many others.
这一技术还可以构建神经网络
This work showcases a neural network
用于测量wifi无线电信号如何
that measures how the wifi radio signals bounce
在房间里传播并且反射出人体姿势
around in the room and reflect off of the human body,
且从这些模糊的信号中
and from these murky waves, it
可估计我们的位置
estimates where we are.
此外 它还可以准确地显示出我们的姿势
Not only that, but it also accurate enough to tell us our pose.
如你所见
As you see here,
因为wifi信号不受光线影响 所以
as the wifi signal also traverses in the dark, this pose estimation works really
在暗景中依然可以精确判断姿势
well in poor lighting conditions.
这是一项了不起的技术
That is a remarkable feat.
但现在 请继续阅读论文
But now, hold on to your papers,
因为没有什么能和你现在所看到的东西
because that’s nothing compared to what you are about to
相比较
see now.
看看这儿
Have a look here.
我们知道wifi信号可以穿过墙
We know that wifi signals go through walls.
也许 这意味着……不可能 对吧
So perhaps, this means that…that can’t betrue, right?
当人进入房间时 它在捕获他的动作
It tracks the pose of this human as he enters the room,
现在 当他消失的时候 看
and now, as he disappears, look,
它仍然知道他在哪里
the algorithm still knows where he is.
没错
That’s right!
这意味着隔着墙它也能检测我们的姿势
This means that it can also detect our posethrough walls!
这是怎样的魔法?现在
What kind of wizardry is that? Now,
注意 该技术并未
note that this technique doesn’t look
检测我们正在观看的视频
at the video feed we are now looking at.
它仅是我们的视觉参考
It is there for us for visual reference.
还要非常注意的是
It is also quite remarkable
它发出的信号比wifi信号要弱一千倍
that the signal being sent out is a thousand times weaker
尽管如此它还可以同时检测多个人的姿势
than an actual wifi signal, and it also can detect multiple humans.
这对彩色图像来说不是什么大问题
This is not much of a problem with color images,
因为我们可以看清每个人
because we can clearly see everyone in an image,
但无线信号
but the radio signals are more
经过多次反射之后更难分辨
difficult to read when they reflect off of multiple
场景中的人物
bodies in the scene.
整个技术通过师生网络结构进行预测
The whole technique work through using a teacher-studentnetwork structure.
老师就是一个标准的
The teacher is a standard pose
姿势预测神经网络 它通过观察彩色图像
estimation neural network that looks at a color image
预测其中的人物姿势
and predicts the pose of the humans therein.
就是这样 没什么新奇的 然而
So far, so good, nothing new here. However,
还有学生网络
there is a student network that looks
观察老师预测的正确结果
at the correct decisions of the teacher, but
但以无线电信号形式输入信息
has the radio signal as an input instead.
结果
As a result,
它能理解不同的无线电信号分布的含义
it will learn what the different radio signal distributions mean and how they
以及信号与人的位置和姿势间的关系
relate to human positions and poses.
顾名思义
As the name says,
老师向学生神经网络显示正确的结果
the teacher shows the student neural network the correct results, and the
学生学习如何用无线电信号来代替图像
student learns how to produce them from radio signals instead of images.
如果十年前有人说正钻研这个项目
If anyone said that they were working on this problem ten years ago,
他们可能
they would have likely
会被送进精神病院 如今
ended up in an asylum. Today,
这成为了现实
it’s reality.
多好的时代!此外
What a time to be alive! Also,
如果你喜欢这一集
if you enjoyed this episode,
请考虑在 patreon网站支持该节目
please consider supporting the show at Patreon.com/twominutepapers.
你能获得很酷的特权
You can pick up really cool perks
例如将您的名字显示为关键支持者
like getting your name shown as a key supporter in the
视频说明也会提到
video description and more.
由于得到您的支持
Because of your support,
我们能够创建这些流畅的视频
we are able to create all of these videos smooth and creamy, in
4k分辨率和每秒60帧支持隐藏式字幕
4k resolution and 60 frames per second andwith closed captions. And,
为节省开支
we are currently saving up
为了能为你呈现更好的视频
for a new video editing rig to make better videos for you.
支持通过PayPal和加密货币付款
We also support one-time payments throughPayPal and the usual cryptocurrencies.
更多有关详细信息
More details about all of these are available
请参阅视频说明 一如既往
in the video description, and as always.
感谢您的关注和支持
Thanks for watching and for your generous support,
下次再见
and I’ll see you next time!

发表评论

译制信息
视频概述

通过电信号,透过墙,观测人体活动的人工智能技术。

听录译者

收集自网络

翻译译者

阿遥

审核员

审核员_AK

视频来源

https://www.youtube.com/watch?v=kBFMsY5ZP0o

相关推荐