感觉谷歌在“把相机对准它了解更多”技术上占据市场已有一段时间了，首先是通过它的翻译应用程序，让你用智能**的相机瞄准外文标牌，在飞行中接收翻译，现在是通过镜头，它扩展了这项技术，为您提供了大量关于您已拍摄（或即将拍摄）照片中对象的信息。

但谷歌的成像人工智能并不是唯一的游戏。今年6月，微软首次推出了针对必应的可视化搜索，这让它大放异彩，你可以通过必应搜索应用（iOS、Android）以及Android上的微软Launcher和Edge进行访问。

你可能不想用多个（或额外的）搜索应用程序或浏览器填满你的**，所以我设计了一系列的实验来看看这两种视觉技术是如何叠加的。

第一轮：日常用品

我在房间里走来走去，随机挑选了各种各样的东西，放在桌子上光线充足的地方。然后，我从大致相同的距离和角度，或者至少从一个有利的角度，拍摄每个物体的照片，这个角度应该足以让每个应用程序都能很好地识别物品。

而且，不，我这次没有用猪。首先，一瓶美味的Framboise lambic：

Google correctly identified the bottle and type of beer, thanks to its excellence in all things text-scanning. A tip of the glass to you, Google. Screenshot: David Murphy

Tapping on that listing provided search results for the beer in question, but no discernible way to place an order from a nearby store and refill my stock. Hmm.

Just for context, here’s how the “item-scanning” feature works in the Bing app. You get a neat little aqua overlay running through your camera’s image that feels a little Star Trek.

Unfortunately, Bing Visual Search couldn’t get anything right about the image, save for the fact that it was indeed a bottle (but not a bottle of wine.)

1 / 4

接下来是一首流行曲：

Google Lens correctly identified the Funko Pop’s game (Diablo III), the fact that it was a Funko Pop, and even provided a link to where one can typically buy said silly items (Hot Topic). Not too shabby. Screenshot: David Murphy

Bing visual search could load up other images from the web of Funko Pops, but mistakenly identified the toy as... Demy. Who is Demy?

1 / 2

一本经典的幻想小说怎么样？

The subject: Terry Goodkind’s classic “Wizard’s First Rule.” Screenshot: David Murphy

Google Lens nailed it. (Ignore the annoying “Screenshot saved” notification at the top.)

Google Lens could also decipher what the book was if you took a photo of its rear, though I was hoping that having the barcode on there would allow Google to suggest places to purchase a new copy. Alas.

Not even close, Visual Search for Bing.

AdvertisementYou can skip ad after 1 second

You can go to the next slide after 1 second

Continue

I’m pretty confident in saying that the Bing tool had no idea what this book was, or possibly that it was even a book to begin with.

1 / 5

为了总结这一类，这里有一款视频游戏，我非常兴奋，但还没有玩很多：

Google Lens is pretty good at scanning all kinds of quirky text, I’ll give it that. Screenshot: David Murphy

...well, almost all kinds of quirky text. I suspect I might have had better luck if I zoomed in on the back of the game’s case, given the text overload.

Well, Bing’s visual search got the sport right, at least.

I didn’t have high hopes.

1 / 4

接下来我们将更实际地使用这些相机技术，下面介绍每个应用程序如何处理一张简单的名片。为了保护隐私，我已经模糊了一些关键细节，但我将在标题中讨论每个应用程序是如何工作的。

As I mentioned earlier, sometimes you just have to zoom in a bit closer in order for an image-recognition app to do anything. From a distance, Google Lens struggled with the business card. Screenshot: David Murphy

Zooming in, however, made it easy to identify all of the card’s key details.

And Google Lens also makes it incredibly easy to add the details from a business card as a new phone contact. Maybe I’ll get around to digitizing my collection one day...

Bing’s Visual Search could identify that the object in question was similar to other photos of white rectangular things.

AdvertisementYou can skip ad after 1 second

You can go to the next slide after 1 second

Continue

After a few more attempts (and zooming in a little), I could at least use Visual Search to call numbers the Bing app identified on the business card itself.

1 / 5

第二轮：纪念碑

既然tl80没有给我这样的实验的旅行预算，“扫描地标和纪念碑以了解更多信息”是每个应用程序的关键功能之一，我只好即兴发挥。我调出纪念碑的照片，用每个应用程序扫描这些照片，这些应用程序还可以扫描你已经在你的相机里拍摄的照片，看看谷歌镜头和必应视觉搜索能识别出什么。

Well, Google Lens got that right, at least. I originally shot the photo in landscape, so maybe a portrait picture of Niagara Falls will work better... Screenshot: David Murphy

Nope.

Nailed it, Bing. You are looking at a photo of a photo of a big-ass waterfall in New York (and Canada)

1 / 3

有人想爬山吗？我听说电缆还不错：

Easy-peasy, Google Lens. Screenshot: David Murphy

No problems from Bing. This is, indeed, Half Dome.

1 / 2

最后，旧金山湾区标志性的旅游陷阱不，不是进进出出的汉堡：

It kind of makes you want to start singing the Full House theme song, doesn’t it?

Everywhere you looooook everywhere you goooooo... Bing has no problem identifying giant bridges.

1 / 2

第三轮：时尚

谷歌镜头（googlelens）和必应（Bing）的视觉搜索都声称能够识别你或你的朋友正在穿的衣服，并建议搭配的商品或你可以购买的商品本身。让我们看看大卫·墨菲收藏的两件作品的效果如何。

Here’s item number one: one of Banana Republic’s finest. Yes, I forgot to steam it. Screenshot: David Murphy

Ah, yes. A... flame resistant shirt. Well, Google Lens did identify other shirts with similar patterns, at least, but I didn’t see any Banana Republic listings specifically in the results.

Although Bing Visual Search didn’t identify the shirt at first...

scrolling down a bit did eventually turn up a Banana Republic listing. So, there’s that. Screenshot: David Murphy

1 / 4

圣诞老人，他的第一个名字。

Google Lens had a little trouble with this one, one of the more unique holiday sweaters I own. Screenshot: David Murphy

Bing’s Visual Search also seemed to struggle at first, but then...

Boom. Here comes House Santa.

1 / 3

判决：谷歌镜头（大部分）完成了任务

我通常发现googlelens是一个更有用的工具，可以在任何特定时刻分析相机中的任何内容。虽然它不是完美的挣扎一点点地标和不那么有趣的时尚应用程序在文本识别和实用性（尤其是当扫描联系信息）。必应善于帮助你找到与照片构图相似的图像，但在找出特定对象方面却不如谷歌镜头，我认为后者的文本识别能力是它的优势所在。

尽管我们大多数人可能会安装（或拉起）谷歌镜头，作为度假时玩玩或在派对上给朋友留下深刻印象的一件事后考虑的事情，但从你的后脑勺移到更中间一点是值得的。我怀疑我是否会走在街上，让谷歌不断地向我提供关于我正在看的东西的信息，但这个应用肯定有它的用途。看到它是如何准确地识别日常物品，我可能会在我的日常旅行中玩它多一点，看看它还能做什么。