经济学人135:计算机模拟视觉

 　　Science and technology

　　科学和技术

　　Computer vision

　　计算机模拟视觉

　　Eye robot

　　你是我的眼

　　Poor eyesight remains¹ one of the main obstacles to letting robots loose among humans.

　　放手让机器人在人类社会自由活动仍存在重大障碍—它们看不清楚。

　　But it is improving, in part by aping natural vision.

　　然而人工视能正在逐渐提高，途径之一就是模拟自然视觉。

　　ROBOTS are getting smarter and more agile² all the time.

　　机器人的反应总是在变得越来越灵活，动作也越来越敏捷。

　　They disarm³ bombs, fly combat missions, put together complicated machines, even play football.

　　它们会拆卸炸弹、驾驶战斗机执行任务、组装复杂机械，甚至还会踢足球。

　　Why, then, one might ask, are they nowhere to be seen, beyond war zones, factories and technology fairs?

　　那么，人们不禁要问，为什么除了在战场、工厂和科技产品展销会，生活中都看不到机器人的踪影呢？

　　One reason is that they themselves cannot see very well.

　　一个原因就是它们自己眼神不大好。

　　And people are understandably wary⁴ of purblind⁵ contraptions bumping into them willy-nilly in the street or at home.

　　机器人跟睁眼瞎差不多，要是把它们弄到大街上，或者摆在家里，搞不好就没头没脑地把人给撞了——对这玩意儿谨慎一点也是人之常情。

　　All that a camera-equipped computer sees is lots of picture elements, or pixels.

　　装有摄像头的计算机能看到的一切，仅仅是大量的图像元素，又称像素。

　　A pixel is merely a number reflecting how much light has hit a particular part of a sensor⁶.

　　像素只不过是一个数值，反映照到传感器某个部位的光线亮度是多少。

　　The challenge has been to devise algorithms that can interpret such numbers as scenes composed of different objects in space.

　　困难在于，要编写出一套计算程序，可以把这些数字再现为空间中不同物体构成的景象。

　　This comes naturally to people and, barring certain optical illusions, takes no time at all as well as precious little conscious effort.

　　这一切对于人类来说，只是一种本能—除非出现某些错觉—立杆而见影，在意识上可谓不费吹灰之力。

　　Yet emulating⁷ this feat⁸ in computers has proved tough.

　　然而事实证明，在计算机上模拟人的这一天赋实非易事。

　　In natural vision, after an image is formed in the retina it is sent to an area at the back of the brain, called the visual cortex, for processing.

　　自然视觉的过程是：视网膜成像后，图像被传送到大脑后部叫做视觉皮层的地方，在那里进行信息处理。

　　The first nerve cells it passes through react only to simple stimuli⁹, such as edges slanting¹⁰ at particular angles.

　　图像经过的第一组神经元只能对简单的视觉刺激作出反射，例如物体朝某些角度倾斜的边缘。

　　They fire up other cells, further into the visual cortex, which react to simple combinations of edges, such as corners.

　　第一组神经元又将兴奋传给视觉皮层更深处的神经元，这些神经细胞可以对简单的物体轮廓作出反应，例如空间中的角落。

　　Cells in each subsequent area discern ever more complex features, with those at the top of the hierarchy¹¹ responding to general categories like animals and faces, and to entire scenes comprising assorted¹² objects.

　　越往后，神经元能识别的图像特征越复杂，最高级别神经区域可以对动物和脸等总体类别作出反应，最后将包罗万象的场景整合到一起。

　　All this takes less than a tenth of a second.

　　而整个过程只需要不到十分之一秒。

　　The outline of this process has been known for years and in the late 1980s Yann LeCun, now at New York University, pioneered an approach to computer vision that tries to mimic¹³ the hierarchical way the visual cortex is wired.

　　很早以前，人们就已经了解这一过程的大致情形。二十世纪80年代末，现就职于纽约大学的雅安?勒存率先涉足计算机视觉研究，试图模拟人脑视觉皮层内神经元层层递进的布线方式。

　　He has been tweaking his convolutional neural¹⁵ networks ever since.

　　从那时起，他就一直在调整和改良他的卷积神经网络。

　　Seeing is believing

　　眼见为实

　　A ConvNet begins by swiping a number of software filters, each several pixels across, over the image, pixel by pixel.

　　卷积神经网络首先用几个软件滤光器，对图像逐一像素地进行扫描，每个滤光器只能通过几个像素。

　　Like the brain's primary visual cortex, these filters look for simple features such as edges.

　　就像大脑的初级视觉皮层，这些滤光器只负责收集物体边缘等简单图像特征。

　　The upshot is a set of feature maps, one for each filter, showing which patches of the original image contain the sought-after element.

　　结果得到一组特征图，每一张特征图对应一个滤光器，显示出原始图像中的哪些块包含要筛选到的要素。

　　A series of transformations¹⁶ is then performed on each map in order to enhance it and improve the contrast.

　　随后，每一张特征图都要进行一系列调整，以提高它的画质、改善它的明暗对比度。

　　Next, the maps are swiped again, but this time rather than stopping at each pixel, the filter takes a snapshot every few pixels.

　　接下来，对这些特征图再次进行扫描，但这一次，滤光器不会在像素上逐一停留，而是每几个像素快拍一次。

　　That produces a new set of maps of lower resolution.

　　这样，得到一组新的分辨率较低的特征图。

　　These highlight the salient features while reining¹⁷ in computing¹⁸ power.

　　这些步骤凸显图像最显著的特征，同时对计算资源进行严格控制。

　　The whole process is then repeated, with several hundred filters probing for more elaborate shapes rather than just a few scouring¹⁹ for simple ones.

　　然后，将整个过程重复一遍，用几百个滤光器探查更为精细的物体形状，而不是随便扫视一些简单的形状。

　　The resulting array of feature maps is run through one final set of filters.

　　由此得到的特征图阵列，被输送经过最后一组滤光器。

　　These classify objects into general categories, such as pedestrians²⁰ or cars.

　　它们可以对物体进行大体归类—是行人还是汽车等等。

　　Many state-of-the-art computer-vision systems work along similar lines.

　　许多尖端计算机视觉模拟系统都采用类似的原理运行。

　　The uniqueness of ConvNets lies in where they get their filters.

　　卷积神经网络的独特之处在于它们的滤光器已经做得登峰造极。

　　Traditionally, these were simply plugged in one by one, in a laborious²¹ manual process that required an expert human eye to tell the machine what features to look for, in future, at each level.

　　以往，滤光器只是一个接一个地接通。这一过程由手工完成，极为繁琐，需要一名专家全程用肉眼观察，然后向机器下达指令，告诉它下一步检索什么样的特征。

　　That made systems which relied on them good at spotting narrow classes of objects but inept²² at discerning anything else.

　　于是，依靠手动操作滤光器的计算机视觉系统，可以识别的物体类别十分有限，而无法分辨其他更多的东西。

　　Dr LeCun's artificial visual cortex, by contrast, lights on the appropriate filters automatically as it is taught to distinguish the different types of object.

　　相比之下，勒存博士的人工视觉皮层，可以在按照设定程序识别不同类型的物体时，自动接通相应的滤光器。

　　When an image is fed into the unprimed system and processed, the chances are it will not, at first, be assigned to the right category.

　　把一张图像输入他的系统进行处理，如果这个系统没有预先存储任何资料，第一次使用时体统有可能会把这张图像错误归类。

　　But, shown the correct answer, the system can work its way back, modifying its own parameters²³ so that the next time it sees a similar image it will respond appropriately.

　　但是，告诉它正确答案之后，系统将重新识别图像，并修改自身的参数，以便下一次再看到类似的图像，可以做出恰当的回应。

　　After enough trial runs, typically 10,000 or more, it makes a decent fist of recognising that class of objects in unlabelled images.

　　经过足够的试运行之后——通常需要进行1万次以上——要在未经标示的图像上识别那一类物体，卷积神经网络可以完成得相当出色。

　　This still requires human input²⁴, though.

　　然而，这个阶段还是需要人类对其进行信息输入。

　　The next stage is unsupervised learning, in which instruction is entirely²⁵ absent.

　　下一个阶段为无监督学习，在这个过程中没有任何人的指导。

　　Instead, the system is shown lots of pictures without being told what they depict²⁶.

　　是的，向勒存的计算机视觉系统展示大量图片，不告诉系统图上画的是什么。

　　It knows it is on to a promising²⁷ filter when the output image resembles the input.

　　如果输出的图像和输入的图像几乎一样，系统就知道自身的滤光器升级了。

　　In a computing sense, resemblance is gauged²⁸ by the extent to which the input image can be recreated from the lower-resolution output.

　　在计算机学上，两张图片是否相像的判断标准是，像素较低的输出图像可以在多大程度上复原为输入的图像。

　　When it can, the filters the system had used to get there are retained.

　　一旦可以还原，为系统所用而达到这种效果的滤光器就被保留下来。

　　In a tribute to nature's nous, the lowest-level filters arrived at in this unaided process are edge-seeking ones, just as in the brain.

　　在这个体现自然界智能的过程中，在无人辅助阶段，滤光器达到的最初等级为物体边缘搜索，正如人脑中的情形一样。

　　The top-level filters are sensitive to all manner of complex shapes.

　　最高等级的滤光器各种光怪陆离的形状都十分敏感。

　　Caltech-101, a database routinely used for vision research, consists of some 10,000 standardised images of 101 types of just such complex shapes, including faces, cars and watches.

　　加州理工101是进行视觉研究常规使用的数据库，它存储了约1万幅标准化图像，描述101类和标准化图像复杂程度相当的物体形状，包括脸、汽车和手表等。

　　When a ConvNet with unsupervised pre-training is shown the images from this database it can learn to recognise the categories more than 70% of the time.

　　当给事先经过无人监督训练的卷积神经网络展示这个数据库中的图像时，它可以通过学习辨认图像的类别，成功几率超过70%。

　　This is just below what top-scoring hand-engineered systems are capable of—and those tend to be much slower.

　　而最先进的手动视觉系统可以做到的也只比这个高一点点—并且它们的辨认速度往往慢得多。

　　This approach which Geoffrey Hinton of the University of Toronto, a doyen of the field, has dubbed²⁹ deep learning need not be confined to computer-vision.

　　勒存的方法多伦多大学的杰弗里?希尔顿是该领域的泰斗，他将这一方法命名为深度学习不一定局限于计算机视觉领域。

　　In theory, it ought to work for any hierarchical system:language processing, for example.

　　理论上，该方法还可以用在任何等级系统当中，譬如语言处理。

　　In that case individual sounds would be low-level features akin¹⁴ to edges, whereas the meanings of conversations would correspond to elaborate scenes.

　　在这种情况下，音素就是语言识别的初级特征，相当于模拟视觉中的物体边缘，而对话的含义则相当于复杂场景。

　　For now, though, ConvNet has proved its mettle³⁰ in the visual domain³¹.

　　然而，目前卷积神经网络已经在视觉领域大显神威。

　　Google has been using it to blot³² out faces and licence plates in its Streetview application.

　　谷歌一直在街道实景应用程序中使用该系统，识别人脸和车牌，对其进行模糊处理。

　　It has also come to the attention of DARPA, the research arm of America's Defence Department.

　　它还引起了美国国防部高等研究计划局的注意。

　　This agency provided Dr LeCun and his team with a small roving robot which, equipped with their system, learned to detect large obstacles from afar and correct its path accordingly—a problem that lesser³³ machines often, as it were, trip over.

　　他们为勒存博士和他的团队提供了一个漫游机器人，给它装上卷积神经网络系统后，这个机器人学会了探测远处的大型障碍物，并相应地纠正行进路线—可以说，没有安装该系统的机器人通常都会在这个问题上绊住。

　　The scooter-sized robot was also rather good at not running into the researchers.

　　这个漫游机器人只有小孩玩的滑板车那么大，却还相当有眼色:它不会撞向研究人员。

　　In a selfless act of scientific bravery, they strode confidently in front of it as it rode towards them at a brisk walking pace, only to see it stop in its tracks and reverse.

　　研究人员们发扬科学家舍身忘我的大无畏精神，做了一个实验：当机器人步履轻盈地向他们开过来时，他们突然昂首阔步迎面冲向机器人。结果发现，机器人半路停下并转向。

　　Such machines may not quite yet be ready to walk the streets alongside people, but the day they can is surely not far off.

　　当然，这类机器人要走上街头与人为伍，或许还略欠火候。但是，它们可以自由行走那一天想必已经不远了。

1 remains   

n.剩余物，残留物；遗体，遗迹

参考例句：

• He ate the remains of food hungrily.他狼吞虎咽地吃剩余的食物。
• The remains of the meal were fed to the dog.残羹剩饭喂狗了。

2 agile   

adj.敏捷的，灵活的

参考例句：

• She is such an agile dancer!她跳起舞来是那么灵巧！
• An acrobat has to be agile.杂技演员必须身手敏捷。

3 disarm   

v.解除武装，回复平常的编制，缓和

参考例句：

• The world has waited 12 years for Iraq to disarm. 全世界等待伊拉克解除武装已有12年之久。
• He has rejected every peaceful opportunity offered to him to disarm.他已经拒绝了所有能和平缴械的机会。

4 wary   

adj.谨慎的，机警的，小心的

参考例句：

• He is wary of telling secrets to others.他谨防向他人泄露秘密。
• Paula frowned,suddenly wary.宝拉皱了皱眉头，突然警惕起来。

5 purblind   

adj.半盲的；愚笨的

参考例句：

• If an administrator has no access to information,it's as if he was purblind and hard of hearing and had a stuffed nose.做管理工作的人没有信息，就是耳目不灵，鼻子不通。
• Even his most purblind supporters knows this is nonsense.即使他最愚蠢的支持者也知道这是无稽之谈。

6 sensor   

n.传感器，探测设备，感觉器（官）

参考例句：

• The temperature sensor is enclosed in a protective well.温度传感器密封在保护套管中。
• He plugged the sensor into a outlet.他把传感器插进电源插座。

7 emulating   

v.与…竞争( emulate的现在分词 )；努力赶上；计算机程序等仿真；模仿

参考例句：

• The possibilities of producing something entirely new by emulating nature's very wide crosses are enticing. 用自然界的非常广泛的杂交方法创造出全新植物种的可能性是诱人的。 来自辞典例句
• The human emulating this archetypal patterning will be quite the accomplished businessperson. 这类原型模式者会是一个很成功的商人。 来自互联网

8 feat   

n.功绩；武艺，技艺；adj.灵巧的，漂亮的，合适的

参考例句：

• Man's first landing on the moon was a feat of great daring.人类首次登月是一个勇敢的壮举。
• He received a medal for his heroic feat.他因其英雄业绩而获得一枚勋章。

9 stimuli   

n.刺激（物）

参考例句：

• It is necessary to curtail or alter normally coexisting stimuli.必需消除或改变正常时并存的刺激。
• My sweat glands also respond to emotional stimuli.我的汗腺对情绪刺激也能产生反应。

10 slanting   

倾斜的，歪斜的

参考例句：

• The rain is driving [slanting] in from the south. 南边潲雨。
• The line is slanting to the left. 这根线向左斜了。

11 hierarchy   

n.等级制度；统治集团，领导层

参考例句：

• There is a rigid hierarchy of power in that country.那个国家有一套严密的权力等级制度。
• She's high up in the management hierarchy.她在管理阶层中地位很高。

12 assorted   

adj.各种各样的，各色俱备的

参考例句：

• There's a bag of assorted sweets on the table.桌子上有一袋什锦糖果。
• He has always assorted with men of his age.他总是与和他年令相仿的人交往。

13 mimic   

v.模仿，戏弄；n.模仿他人言行的人

参考例句：

• A parrot can mimic a person's voice.鹦鹉能学人的声音。
• He used to mimic speech peculiarities of another.他过去总是模仿别人讲话的特点。

14 akin   

adj.同族的，类似的

参考例句：

• She painted flowers and birds pictures akin to those of earlier feminine painters.她画一些同早期女画家类似的花鸟画。
• Listening to his life story is akin to reading a good adventure novel.听他的人生故事犹如阅读一本精彩的冒险小说。

15 neural   

adj.神经的，神经系统的

参考例句：

• The neural network can preferably solve the non- linear problem.利用神经网络建模可以较好地解决非线性问题。
• The information transmission in neural system depends on neurotransmitters.信息传递的神经途径有赖于神经递质。

16 transformations   

n.变化( transformation的名词复数 )；转换；转换；变换

参考例句：

• Energy transformations go on constantly, all about us. 在我们周围，能量始终在不停地转换着。 来自辞典例句
• On the average, such transformations balance out. 平均起来，这种转化可以互相抵消。 来自辞典例句

17 reining   

勒缰绳使（马）停步( rein的现在分词 )； 驾驭； 严格控制； 加强管理

参考例句：

• "That's a fine bevy, Ma'm,'said Gerald gallantly, reining his horse alongside the carriage. "太太！好一窝漂亮的云雀呀！" 杰拉尔德殷勤地说，一面让自己的马告近塔尔顿的马车。
• I was a temperamental genius in need of reining in by stabler personalities. 我是个需要由更稳重的人降服住的神经质的天才。

18 computing   

n.计算

参考例句：

• to work in computing 从事信息处理
• Back in the dark ages of computing, in about 1980, they started a software company. 早在计算机尚未普及的时代（约1980年），他们就创办了软件公司。

19 scouring   

擦[洗]净，冲刷，洗涤

参考例句：

• The police are scouring the countryside for the escaped prisoners. 警察正在搜索整个乡村以捉拿逃犯。
• This is called the scouring train in wool processing. 这被称为羊毛加工中的洗涤系列。

20 pedestrians   

n.步行者( pedestrian的名词复数 )

参考例句：

• Several pedestrians had come to grief on the icy pavement. 几个行人在结冰的人行道上滑倒了。 来自《简明英汉词典》
• Pedestrians keep to the sidewalk [footpath]! 行人走便道。 来自《现代汉英综合大词典》

21 laborious   

adj.吃力的，努力的，不流畅

参考例句：

• They had the laborious task of cutting down the huge tree.他们接受了伐大树的艰苦工作。
• Ants and bees are laborious insects.蚂蚁与蜜蜂是勤劳的昆虫。

22 inept   

adj.不恰当的，荒谬的，拙劣的

参考例句：

• Whan an inept remark to make on such a formal occasion.在如此正式的场合，怎么说这样不恰当的话。
• He's quite inept at tennis.他打网球太笨。

23 parameters   

因素，特征； 界限； （限定性的）因素( parameter的名词复数 )； 参量； 参项； 决定因素

参考例句：

• We have to work within the parameters of time. 我们的工作受时间所限。
• See parameters.cpp for a compilable example. This is part of the Spirit distribution. 可编译例子见parameters.cpp.这是Spirit分发包的组成部分。

24 input   

n.输入(物)；投入；vt.把(数据等)输入计算机

参考例句：

• I will forever be grateful for his considerable input.我将永远感激他的大量投入。
• All this information had to be input onto the computer.所有这些信息都必须输入计算机。

25 entirely   

ad.全部地，完整地；完全地，彻底地

参考例句：

• The fire was entirely caused by their neglect of duty. 那场火灾完全是由于他们失职而引起的。
• His life was entirely given up to the educational work. 他的一生统统献给了教育工作。

26 depict   

vt.描画，描绘；描写，描述

参考例句：

• I don't care to see plays or films that depict murders or violence.我不喜欢看描写谋杀或暴力的戏剧或电影。
• Children's books often depict farmyard animals as gentle,lovable creatures.儿童图书常常把农场的动物描写得温和而可爱。

27 promising   

adj.有希望的，有前途的

参考例句：

• The results of the experiments are very promising.实验的结果充满了希望。
• We're trying to bring along one or two promising young swimmers.我们正设法培养出一两名有前途的年轻游泳选手。

28 gauged   

adj.校准的；标准的；量规的；量计的v.（用仪器）测量( gauge的过去式和过去分词 )；估计；计量；划分

参考例句：

• He picked up the calipers and gauged carefully. 他拿起卡钳仔细测量。 来自《现代汉英综合大词典》
• Distance is gauged by journey time rather than miles. 距离以行程时间而非英里数来计算。 来自辞典例句

29 dubbed   

v.给…起绰号( dub的过去式和过去分词 )；把…称为；配音；复制

参考例句：

• Mathematics was once dubbed the handmaiden of the sciences. 数学曾一度被视为各门科学的基础。
• Is the movie dubbed or does it have subtitles? 这部电影是配音的还是打字幕的？ 来自《简明英汉词典》

30 mettle   

n.勇气，精神

参考例句：

• When the seas are in turmoil,heroes are on their mettle.沧海横流，方显出英雄本色。
• Each and every one of these soldiers has proved his mettle.这些战士个个都是好样的。

31 domain   

n.(活动等)领域，范围；领地，势力范围

参考例句：

• This information should be in the public domain.这一消息应该为公众所知。
• This question comes into the domain of philosophy.这一问题属于哲学范畴。

32 blot   

vt.弄脏（用吸墨纸）吸干；n.污点，污渍

参考例句：

• That new factory is a blot on the landscape.那新建的工厂破坏了此地的景色。
• The crime he committed is a blot on his record.他犯的罪是他的履历中的一个污点。

33 lesser   

adj.次要的，较小的；adv.较小地，较少地

参考例句：

• Kept some of the lesser players out.不让那些次要的球员参加联赛。
• She has also been affected,but to a lesser degree.她也受到波及，但程度较轻。