Now, artificial intelligence powers fashion ecommerce in India

如今，人工智能助力印度时尚电子商务

You see a dress you like on a fashionportal and want to compare it with similar ones to make a choice. It seemssimple. But do you realize how smart your brain has to be to decipher that onedress is similar to another? Color, light, dimensions, depth, gender – yourbrain processes a multitude of inputs to make the call.

当你在一个时尚门户网站上看到一个喜欢的裙子时，并想与其他相似的裙子比较来做出选择。这看似很简单，但是你是否意识到你的大脑需要多么智能，才能够辨识一个裙子与另一个裙子很相似呢？颜色，光亮，尺寸，深度，性别，你的大脑需要处理大量的信息来做出回应。

Now, imagine trying to get a machineto do that. To come up with intelligent suggestions for dresses that match theone you like, a computer has to convert images into numerical and symbolicdata. This is a complex task involving geometry, probabilities, learningtheory, and so on. Of course, once it learns to do that, it can scan a lot moredresses much quicker than we humans can, so we get better choices.

现在，幻想一下尝试用机器来做这些事情。利用智能建议来搭配你喜欢的裙子，计算机需要将图像转换成数字和符号数据。这是一项包含几何，概率以及学习理论等的复杂工作。当然，一旦机器学会了这些，与我们人类相比，它就能以更快的速度扫描更多的裙子，这样我们就能得到更好的选择。

Taggingto seeing

通过标签来观察

Indian fashion portalsVoonikandCraftsvillaare rolling this out today: if you goto these sites and click on a Bollywood saree or terracotta jewelry you like,artificial intelligence will kick in and recommend similar sarees and jewelryto you instantly. (For a couple of weeks, the feature will be in A/B testingmode, which means only half the users can access it.)

印度时尚门户Voonik以及Craftsvilla今天推出这项服务：如果你访问这些网站并点击你喜欢的宝莱坞纱丽或赤土首饰，人工智能就会弹出并立刻为你推荐相似的纱丽或珠宝。（在未来的几周，这一功能将会处于A/B测试模式，这意味着只有一半的用户可以进入）

We’ve got used to seeingrecommendations for products when we click on books, apparel, or anything elseonecommercesites. But those are different from visual comparison.“Earlier, we only had recommendations based on tags. We look at the physicalattributes of items (color, sleeve, fabric, etc.) and suggest items which havesimilar attributes. With the new feature, we show items that are visuallysimilar to the item the user is viewing,” explains Sujayath Ali, co-founder andCEO of Voonik.

我们已经习惯于在电子商务网站上点击书籍，器械或其他东西时看到一些产品的推荐。但是这与视觉比较还有很大的不同。“早期，我们只有基于标签的推荐。我们看到条目的相关物理属性（颜色，袖套，面料等），并建议有相似属性的条目。有了新功能后，我们就能推荐与用户浏览的外观相似的条目，”Voonik的联合创始人以及CEO，Sujayath Ali解释道。

Tags are text-based descriptors whichmay even include visual attributes like color. But they don’t tell you theexact shade of color, pattern, or shape of the dress like computer vision can.That’s why you often find oddball recommended products. After all, describingan object with tags cannot mimic how we see an object in its entirety. Butcomputer vision comes close to seeing an object as the human brain does.

标签是基于文本的描述符，其中还可能会包括可视化的特征，比如颜色。但是他们并不能像计算机视觉那样告诉你确切的色度，图案或者裙子的形状。这就是为什么你经常会看到一些古怪的推荐的原因。总而言之，用标签来描述对象并不能完全的模拟我们所看到的物体。但是计算机视觉可以类似于人类大脑一样观察物体。

Apart from tags, recommender systemson ecommerce sites rely on user data – such as, what you’ve liked earlier orwhat’s popular among other users. But for fashion portals like Voonik andCraftsvilla, in particular, it makes sense for the recommendations of similarproducts to be based on the visual attributes of something you click.

除了标签之外，电子商务网站的推荐系统也依赖于用户数据，比如说，你之前喜欢什么以及在其他用户之中比较流行什么。但特别是像Voonik以及Craftsvilla这类时尚门户网站，主要是基于用户点击物品的可视化特征，从而做出有效的产品推荐。

Behind this visual recommendationengine is a cool product called MADstack – a cloud-based suite of computervision and AI modules that an ecommerce or any other site can plug into with anAPI(applicationprogram interface) to create new experiences for customers.

在可视化推荐引擎背后一个很酷的产品叫做MADstack，这是一种基于云技术的计算机视觉以及人工智能模块套装，电子商务网站或其他网站可以接入API（应用程序接口），为客户创造新的体验。

MAD stands for Mind-Abled Devices,and MADstack is a creation ofMad Street Den, an Indianstartupthat’s barely two years old. Its founders, however, areseasoned pros from the hallowed domains of AI, user interface, and neuroscienceinSilicon Valley.

MAD代表思维可用设备，而MADstack也是Mad Street Den公司的产品，这是一个仅仅成立两年的创业公司。然而它的创始人是来自于硅谷神圣领域的人工智能，用户界面，以及神经科学方面的资深专家。

Ashwini Asokan led the mobileinnovation team at Intel’s Interaction and Experience Research Lab. Herhusband, and Mad Street Den co-founder, Anand Chandrasekaran was part of an AIteam atStanfordUniversitythat built Neurogrid – adevice that simulates the neurons and synapses of the brain to control bioniclimbs.

Well, now they’re here to help youpick a Bollywood saree.

Ashwini Asokan在英特尔交互以及体验研究实验室领导移动创新团队。她的丈夫Anand Chandrasekaran，也是Mad Street Den的联合创始人，Anand Chandrasekaran是斯坦福大学人工智能团队的成员之一，他们打造了可以模拟大脑神经元和突触来控制仿生肢的设备—Neurogrid。

Morethan visual search

不仅仅是视觉搜索

WhenTech in Asiafirst

profiled the couple last year, right after they moved to theirhometown of Chennai inIndia,they were experimenting with fun ways to use computer vision in everyday life,games, and business.

刚好在他们搬到在印度的家乡Chennai后，Tech in Asia网站去年第一次介绍了这对夫妇，他们试验用有趣的方式在日常生活，游戏以及商务中使用计算机视觉技术。

Amazon’s Firefly, billed as “visualsearch on steroids,” had come out around then. But, of course, the Fireflywould only work with Amazon. What the ‘Mad’ couple decided to do was to putthat capability and more on the cloud, for any ecommerce site to use.

标榜为“视觉搜索激素”的亚马逊萤火虫当时已经出现。但是萤火虫只能在亚马逊网站上使用。这对“疯狂”的夫妇决定将此功能扩展到云技术，这样任何电子商务网站都可以使用。

Othervisual search playershave entered the ecommercescene inIndiasince then.Wazzat Labs, which recently graduatedfrom the TargetacceleratorinBangalore,has gone live on the US-based retailer’s website.iLenze, incubated at TLabs, is a new entrant. And last week, Flipkartannounced a visual search feature from Singaporean startup Visenze, even though it’s stillat a beta stage of development.

从那以后其他视觉搜索的玩家就开始进军印度电子商务行业。最近从位于班加罗尔Targetaccelerator试验成功的Wazzat Labs，已经在TLabs孵化的美国零售商网站行业新军.iLenze上使用。上周Flipkart从新加坡创业公司Visenze发布了新的视觉搜索功能，尽管它仍处于开发测试阶段。

So, visual search has suddenly becomethe buzzword in Indian ecommerce.

Mad Street Den co-founder AshwiniAsokan points out, however, that visual search-based recommendation is only oneof the tricks in MADstack’s bag. While object recognition powers the visualsearch, computer vision offers other possibilities such as recognizing thegestures and emotions of customers for businesses to understand user experienceand respond accordingly. Mad Street Den is thus positioned as a holisticartificial intelligence company.

因此，视觉搜索忽然成为了印度电子商务届的流行语。

Mad Street Den的联合创始人Ashwini Asokan指出，虽然基于视觉搜索的推荐仅仅是MADstack服务包的招数之一。然而对象识别功能支援了视觉搜索，计算机视觉提供了其他的功能，比如说识别客户的手势以及情感，这样企业就会理解客户的体验并相应地做出回应。Mad Street Den也因此被定为为全面的人工智能公司。

Newlevel of personalization

定制化新水平

“Our platform is architected in sucha way that we can scale up any kind of computer vision feature,” Asokan tellsTech in Asia. “We start with visual recommendation, but as we goalong our technology is such that we move towards visual personalization.”

“我们以这种方式打造平台，这样我们就可以扩展任何类型的计算机视觉功能，”Asokan告诉Tech in Asia，“我们从视觉推荐开始，但是当推广我们的技术时，就开始转向视觉定制化。”

For example, if I click on a pinkBollywood saree, MADstack can see that it’s a fuchsia shade I prefer and notbright pink, as well as the patterns that appeal to me. This lends itself tomore granular personalization of my shopping experience than anything atext-based search can provide.

例如，如果我点击一个粉红色的宝莱坞纱丽，MADstack就会明白我喜欢的是樱红色，而不是亮粉色，以及吸引我的图案。这可以为我提供更加完美的定制化购物体验，这是其他任何基于文本的搜索不能提供的。

Visual search is still a work inprogress, even though the technology has been around for years.Google acquired a visual search company –Like.com – way back in 2010. And yet, five years later the technology is stillnot widespread in ecommerce or other use cases.

视觉搜索仍然处于试运行阶段，尽管这项技术已经存在多年。谷歌在2010年收购了一家视觉搜索公司——Like.com。然而五年后这项技术仍然没有在电子商务或其他应用案例中普及开来。

“People still have to work out how tomake it usable,” says Asokan. “Visual search is a fundamental change in the waysomebody searches for something as people are used to text search. Nobody takespictures of things and searches for them.”

“人们仍然在努力研究如何使其可用，”Asokan说。“视觉搜索是对人们习惯使用的文本搜索方式的根本性转变。没有人去留意并搜索他们。”

‘Clicks’take on a new meaning

‘点击’有新的含义

Mad Street Den itself started outwith recognition of gestures and facial expressions captured on smartphonecameras, when it built the MADstack last year. It didn’t take long to extendthat to visual search and recommendation, which had more takers in the market.But then to get it market-ready took five months afterraising US$1.5 million in funding.

在去年打造MADstack时，Mad Street Den本身从智能手机摄像头识别手势以及面部表情开始。它并没有花费很长时间就将其延伸到视觉搜素以及推荐服务中，这在市场上有更多的接受者。然而在筹集150万资金后，花费了5个月的时间让其充分投入市场。

Now the company is all set to rollout not only visual recommendation on ecommerce sites,but other possibilities of artificial intelligence as well.

现在公司不仅在电子商务网站上推出了可视化推荐功能，还可能会推出其他人工智能服务。

For now, Craftsvilla and Voonik arethe first of several ecommerce sites which will be plugging into the MADstack,Asokan tellsTech in Asia. After starting withvisual recommendations, several of these sites will roll out visual search –which will let you click a picture of somebody wearing a dress you like andsearch for it online, as demonstrated in this video.

Asokan告诉Tech in Asia，目前Craftsvilla和Voonik是最先嵌入MADstack的网站之一。在推出可视化推荐后，很多网站还将会推出可视化搜索，正如视频中表明的那样，这可以让你点击一张穿着你喜欢的裙子的人物图片，并在网上进行搜索。

声明：本文为个人翻译最新商务类新闻文章，均有中英文对照，转载请与作者联系，尊重版权！