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

从包含深度信息的图像中人工智能学习几何描述 – 译学馆
未登陆,请登陆后再发表信息
最新评论 (0)
播放视频

从包含深度信息的图像中人工智能学习几何描述

AI Learns Geometric Descriptors From Depth Images | Two Minute Papers

亲爱的学霸同学们 这里是Károly Zsolnai-Fehér带来的两分钟论文
Dear Fellow Scholars, this is Two Minute Papers with Károly Zsolnai-Fehér.
今天 我们要来讨论一篇很棒的论文 它给我们展示了
Today, we’re going to discuss a great piece of work that shows us
近来基于神经网络的技术是多么的高效和通用
how efficient and versatile neural network-based techniques had become recently.
在这里 输入一些RGB-D图片
Here, the input is a bunch of RGB-D images,
即携带深度信息的照片
which are photographs endowed with depth information,
而我们得到的输出会是一个场景的全三维重建
and the output can be a full 3D reconstruction of a scene,
以及很多很多其他 这些我们待会将看到
and much, much more, which we’ll see in a moment.
这项工作一般是由手工创造的描述符来完成
This task is typically taken care of by handcrafting descriptors.
描述符是对图像或是其他数据结构做有用处理的一种专门的表述
A descriptor is a specialized representation for doing useful tasks on images and other data structures.
举例来说 如果我们想要建立一个识别黑白图片的算法
For instance, if we seek to build an algorithm to recognize black and white images, a useful
那么其中一个有用的描述符 将必然包含图中可见颜色的总数
descriptor would definitely contain the number of colors that are visible in an image,
以及所有这些颜色的列表
and a list of these colors.
再重申一次 几十年来 这些描述符通常都是由科学家们手工创造的
Again, these descriptors have been typically handcrafted by scientists for decades.
所以 新的问题就需要新的描述符 于是就有了新的论文
New problem, new descriptors, new papers.
但是这次不一样了 因为通过一种学习算法、卷积神经网络和siamese网络
But not this time, because here, super effective descriptors are proposed automatically via
而自动生成了超级有效的描述符
a learning algorithm, a convolutional neural network and siamese networks.
真是让人难以置信!
This is incredible!
过去 为了创造这样的描述符 对于一个具体的问题 通常需要极聪明的研究人员工作好几年
Creating such descriptors took extremely smart researchers and years of work on a specific
然而 其结果还不如现在这些好
problem, and were still often not as good as these ones.
顺便说一下 我们在之前的一期节目里讨论过siamese网络
By the way, we have discussed siamese networks in an earlier episode, as always, the link
节目链接可以在视频描述栏里找到
is available in the video description.
和你想的一样 好几种非常酷的应用都由此产生
And as you can imagine, several really cool applications emerge from this.
第一个例子 将此技术和用来寻找噪声测量数据规律的RANSAC技术相结合
One, when combined with RANSAC, a technique used to find order in noisy measurement data,
可以做到只需少量图片就可重建三维场景
it is able to perform 3D scene reconstruction from just a few images.
并且秒杀其他竞争对手
And it completely smokes the competition.
第二个例子 利用边框做形态评估
Two, pose estimation with bounding boxes.
只需提供一个物体样品 那么在一个有此物体和其他物体相堆叠的场景中
Given a sample of an object, the algorithm is able to recognize not only the shape itself,
这个算法不仅能识别出其形状 还能识别其摆放方向
but also its orientation when given a scene cluttered with other objects.
第三个例子 可实现对应搜索
Three, correspondance search is possible.
这真是太酷了!
This is really cool!
这意味着可以在不同物体上识别一个语义相似的几何形状
This means that a semantically similar piece of geometry is recognized on different objects.
比如说 此算法可以学习“把手”这个概念 然后就可以识别
For instance, the algorithm can learn the concept of a handle, and recognize the handles
在各种不同物体上的把手 例如摩托车、推车、椅子等等
on a variety of objects, such as on motorcycles, carriages, chairs, and more!
此项目的源程序可供下载
The source code of this project is also available.
好棒!
Yoohoo!
神经网络现已快速地在很多研究领域占据主导地位
Neural networks are rapidly establishing supremacy in a number of research fields,
我很高兴可以亲眼见证这一让人难以置信的研究进展
and I am so happy to be alive in this age of incredible research progress.
记住订阅本系列 然后点击铃铛图标
Make sure to subscribe to the series and click the bell icon, some amazing works are coming
接下来几期仍有精彩论文奉上 并且会更有趣
up in the next few episodes, and there will be lots of fun to be had.
感谢收看和支持 下次再见!
Thanks for watching and for your generous support, and I’ll see you next time!

发表评论

译制信息
视频概述

本集讨论了一种新的基于神经网络的技术,可用来自动生成描述符,及利用此技术的几种应用

听录译者

收集自网络

翻译译者

豆子

审核员

知易行难

视频来源

https://www.youtube.com/watch?v=1U3YKnuMS7g

相关推荐