How to find an image on the Internet from a photo - Google, Yandex search by photo. Convenient image recognition function with Google search by image Photo recognition service

Sometimes you need to find duplicate photos. For example, a beautiful girl posted a photo on the Internet, but you are not sure that it is she in the picture. In this case, you may need to search the photo on the Internet. If the network is full of such photos, then it is clearly fake. Special services and programs have been created for this purpose. Let's look at some of them.

How to find duplicate photos on the Internet

So, you have a photo and you want to find all its duplicates on the Internet. Use services from Yandex or Google, or separate programs.

Google

The algorithm of actions is as follows:

Follow this link .
In the search bar, click on the camera thumbnail.

Provide a link to the desired photo or upload a photo from your PC.

Click on the link "All sizes".

Now the sites where this image appears will be displayed.

Yandex

Yandex also has a service similar to the previous one:

Teeneye

Another service for searching by image is Tinay. Along with duplicate photos, he also finds their components. Tineye has its own, largest database of indexed images. It has everything, and this is not sarcasm. It's easy to use:

FindFace

In addition to online services, there is a special program for searching photos on the Internet. FindFace searches for people's pages on the VKontakte social network that have a similar photo. It works online, that is, it does not require installation on a computer:

To my surprise, she actually selected 30 pages with faces as close as possible to the original. However, FindFace only searches for people in Russia.

By the way, you can download the FindFace application for Android on Google Play and use it from your phone.

PhotoTracker Lite

To quickly search for similar images using services from Yandex, Google and Tinay, install the PhotoTracker Lite browser extension. It works with almost all modern browsers. You can find it in the Chrome extension store.

After installation, you need to right-click on any picture on the Internet and select “Find this image”.

Conclusion

Now you know how to find duplicate photos of people, objects and other objects on the Internet.

There aren't many augmented reality apps in the App Store. And there are even fewer of them that are truly high-quality and interesting. For the most part, these are games that cannot be called serious projects. Nevertheless, sometimes developers release projects worth attention. For example, an interesting translator Word Lens, which can translate text using your smartphone’s camera on the fly. What about identifying objects from photographs?

Camfind is an unsightly app that 90% of users would easily skip when searching for something new in the Apple app store. The name, icon and description are designed in such a way that the first thought that comes to mind is “another scam”. However, looking at the “price” and the lack of in-app purchases, curiosity takes over. That's how I downloaded this app.

Surprisingly, the stated features actually work, but with minor reservations. Camfind is an app that uses a photo you take to identify the subject in it. Let's say you pass by a store and notice some product on the window. You take a photo, which is uploaded to the developer’s servers and then recognized. After identification, you are transferred to the Google search engine to view the results. Thus, you can find out about any object unknown to you simply by photographing it.

In addition, there are more standard identification methods: scanning barcodes and QR codes, voice recognition - but all this has long been implemented in other applications, which is why recognizing objects in photographs is the most interesting. I took a few photos and found that the app did a pretty good job with each one. Based on the photo of the TV, Camfind gave me the result not only that it was a color, flat-panel TV, but also its brand. After taking a photo of a flower in the hallway, the application gave me “a green plant in an indoor pot.”

According to the developers, you can also take pictures of various event posters or some places in the city to get additional information about them. However, apparently not for this country. In addition, a suspicion creeps in that your photo is not processed by some specialized algorithm, but by ordinary people who sit and “sort out” the photos sent to them (perhaps not all, but some part).

In any case, as with the Word Lens translator, the application does not serve any specific purpose of making your life easier. At the moment, this is more of a technology demonstration than a final product. It is quite possible that in the future such developments will be developed into much more useful applications.

However, the application is entirely free, so you can download it and see for yourself how it works. At the very least, it's interesting.

Hello, dear readers of the blog site. Until recently, search engines could only search by words. No, of course, there were such, but until a certain period they operated only with the words that the user entered in the search bar.

How does searching by uploaded images differ from regular search?

The search engine did not understand what exactly was shown in the photos that it returned in the results, but was only guided by the words that appeared in the text next to this image (on the pages of the sites where it found them) or that were written in its attributes. In fact, for the search engine, the picture was a “black box”, which he could judge only by indirect signs (its description).

As a result, the query “blue chicken” could show “pink elephants”. Of course, such errors in photo searches were corrected manually (so-called), but this was done only for the most frequently entered queries. And that's not the main thing.

The important thing is that it was impossible show the search engine an image of poor quality, so that he would find you the original in high resolution or show him a photo of a person (read), and he would tell you that this is such and such an artist, poet, musician or just a person described on at least one page on the Internet .

Sometimes a photo may be one of many in a series (photo report, photo instruction, different angles) and you may want to find all the other images from the same series to understand the essence. How to do it? What words should you type into the search bar? Here's another example. You saw a sofa in a photo and wanted to know where exactly the same one was sold and at what price.

A difficult task, or even unsolvable (to a first approximation), if the query is entered in words. Here you need to somehow upload the picture to the search engine and the latter must understand what exactly is depicted on it and try to give you answers to all the questions posed just above.

Until recently, image search could not do this, but now it can. Search by sample image(photos or any other image) are now supported by both Google and Yandex. Moreover, the latter learned to do this only quite recently, but nevertheless learned.

If you try to dive into the full depth of the principles of this action, then most of us are unlikely to find it interesting. In this regard, I really liked the explanation of the Yandex representative.

The picture is divided into small fragments, which can be called virtual words. Well, then the process follows the same logic as a regular search. The same set of visual words is searched, and the closer it is to the user's uploaded image, the higher it will appear in the search results.

How does Google photo search work?

Let's look at all this using the examples of Yandex and Google. Let's start with the largest search engine in the world. In order to get into the holy of holies, you can click on the “Pictures” button on the regular search page, or you can immediately:

Click on the camera icon located in the right area of the Google search bar. You are given two options to upload the desired picture or photo to the search- indicate its address (it can be copied by right-clicking on the photo and selecting the “copy image URL” option, or something similar in meaning) on the Internet or upload it from your computer.

In the example shown, I simply indicated the URL of the image that I found on the official Wikipedia website ().

In the search results I got this picture:

Google told me that the photo shows Albert Einstein in his youth, and also suggested that I look at the same image, but in a larger or smaller size. You can also look at similar pictures, and under them you can see the web pages where this graphic file takes place.

In the search bar, you can enter clarifying words, for example, if you want to find out about all the vicissitudes of this person’s fate, then enter the word “biography”. As a result, pages will be found on which the search found the picture you uploaded and on which there is a biography of the person depicted in it.

I mentioned two main ways to upload an image to Google search - provide a link to the graphic file or upload it from your computer. But there is also a third way to activate this action.

If you are in Google Chrome, then simply move your mouse to the picture or photograph on the page that interests you for one reason or another (for example, you want to find out whether your virtual friend posted a true photo on VKontakte or whether it was captured by some famous person).

As a result, a context menu will appear and you will need to select “Find this image in Google” from it. After this, you will find yourself in the already familiar Google image search window with all the collected information about your friend’s photo.

Search by image files in Yandex

Quite recently, the leader of the RuNet search market acquired a similar tool. They called the new technology “Computer Vision” and gave it the code name “Siberia”. In this case, the picture is divided into visual words (areas of contrast change, borders, etc.) and the entire database available in Yandex is searched for the presence of this set of visual words in other images.

And only then those in which these visual words are in the same order as in the original uploaded picture are selected. In practice, this action looks very similar to Google - in the right search area for photos from Yandex the camera icon is located, which you will need to click to download the graphic file.

Although, if you have the URL of the desired image, you can paste it directly into the graphical search bar and click on the “Find” button, as shown in the previous screenshot.

You can find out the URL of an image on a web page by right-clicking on it and selecting the context menu item “Copy image address” or similar (different names are used in different browsers).

If you need to upload a picture to the search from your computer, then click on the camera icon.

The search results will look something like this:

As you can see, from them you can also quite easily understand that the photo shows the great Einstein, but on Google I somehow liked the design of the results more. It is possible that Yandex will still need to work on this. Those found can also be sorted by size and type.

In general, in most cases the services described should be enough for you, but it is possible that Tinay you will need it someday. Here, again, you are offered a choice of two ways to download a sample image - by entering the URL address or directly from your computer.

This service is well suited for searching for similar photos or individual components from which it was assembled. As you can see from the screenshot below, Tinay found the sources of the original images that made up the sample.

Chinese search engine Taobao has the ability to search for products based on the photo you uploaded. As a result, you will not only receive the addresses of those online stores where you can buy it, but you will also be able to choose the most profitable offer for you.

True, to work with it you need to know Chinese. However, there are several sites that allow you to search for products by photo through the Taobao database, but at the same time they have a Russian-language interface and the search results are also translated into Russian.

The results of searching for a product by photo look like this:

Good luck to you! See you soon on the pages of the blog site

Search by photo in Google

Provide a link to the image on the Internet
Uploading a file from a computer

As a result, we get a complete list of similar pictures from the image that was chosen as a sample:

There is another good way that works in the Chrome browser. While on the page with the image you are interested in, move the mouse cursor to it, right-click and in the tooltip that opens, select “Find image (Google)”:

You will be immediately taken to the search results page!

Search by images in Yandex

With Yandex, everything is no less simple than with Google :) Follow the link https://yandex.by/images/ and click the camera icon in the upper right corner:

Enter the address of the image on the Internet or upload it from your computer (you can simply drag it to a special area at the top of the browser window):

The search result looks like this:

You instantly have access to the following information:

What are the online dimensions of the image you uploaded as a sample for searching?
List of sites where it appears
Similar pictures (modified based on the original one or based on which the algorithm decided on their semantic similarity)

Many have probably already heard about the online service TinEye, which Russian-speaking users often call Tinai. It is developed by experts in the field of machine learning and object recognition. As a consequence of all this, Tinay is great not only for finding similar pictures and photographs, but their components.

TinEye's indexed image database contains more than 10 billion items, and is the largest in the entire Internet. “Everything can be found here” - this phrase perfectly characterizes the service.

There is another way to search in one click. By default, the “Show quick search icon” item is activated in the application settings. When you point at a photo or picture, a round green icon pops up, clicking on which starts a search for similar images - search results for Google, Yandex, Tinay and Bing will automatically open in new tabs.

The extension was created by our compatriot, whose hobbies are closely related to photography. He originally created this tool to quickly find his photos on other people's sites.

When you might need it

You are a photographer, you post your photos on the Internet and want to see on which sites they are used and where your copyrights may be violated.
You are a blogger or copywriter, write articles and want to choose an “unhackneyed” image for your material.
What if someone uses your photo from your VKontakte or Facebook profile as an avatar on a forum or a fake account on some social network? But this is more than possible!
You found a photo of an actor you know and want to remember his name.

In fact, there are a huge number of cases when searching by photo can be useful. You can give another example...

How to find the original of a given image

For example, you have some kind of photograph, perhaps cropped or photoshopped, and you want to find its original, or a version in better quality. How to do it? Conduct a search in Yandex and Google, as described above, or using PhotoTracker Lite and get a list of all images found. Next, follow the following:

The original image is usually larger and of better quality than the altered copy resulting from cropping. Of course, you can set a picture to any size in Photoshop, but when you enlarge it relative to the original, artifacts will always be observed. They can be easily noticed even with a cursory visual inspection.
Original photographs often have watermarks indicating the author of the photo (last name, website address, company name, etc.). Of course, anyone can add a watermark to absolutely any image, but in this case, you can look for a sample photo on the website or by the author’s last name, he probably posts his portfolio online somewhere.
And finally, a very simple sign. If your sample photo is black and white (sepia, etc.), and you find the same, but full-color photo, then you obviously do not have the original. much more difficult than converting a color photo to black and white :)

How far has technology come?

If a person looks at a photo where a woman with a book in her hands next to a dog is watching a UFO from the window, then he will most likely determine the age and mood of the lady, the breed of the animal, what is written in the book and in what font.

In 2016, a computer can do the same! And unlike a person, he will also determine the location of the shooting and point out traces of the “sticker” of the flying saucer.

And you can try to do all this online, for free and right now!

1. Looking for traces of photomontage

Go to 29a.ch/photo-forensics and use the button Open File upload a photo.

And we begin to explore it with the tools on the right panel. Using a magnifying glass, we try to find poorly disguised traces of processing:

Layer analysis allows you to detect “patches”:

Didn't you use a clone brush here:

You can learn more about how to find traces of editing in a photo in a training video from the creators of the service:

2. Read the metadata

Go to Jeffry’s Exif Viewer and upload an image (raw, taken on a digital camera or phone). In a second we know on what device and at what time it was taken, and we also see the location of the shooting on the map:

3. Recognize a celebrity in a photo

Check out beface.com to find out which celebrity is in the photo. Download the file and wait a little less than a minute:

Click on the button Search Wikipedia. A list of faces similar to the one uploaded will soon appear:

To find out the name of the person in the photo, you need to hover your mouse over him. My experience has shown that the service very accurately identifies famous people.

4. Determine the font of the inscription

In a special section of myfonts.com you can upload a picture with an inscription and determine the font in which it was made. The result page displays five sample inscriptions of maximum similar fonts (with the same text as the downloaded one):

5. Recognize text

Newocr.com is for those who want high-quality text recognition, but do not want to pay for FineReader. Upload the scan, wait a while and enjoy the result:

The finished text can be downloaded, copied to the clipboard, or translated using Google Translate.

The service has many analogues, but based on my experience, newocr.com does the best job of recognizing the Russian language.

You can use Google to search for similar photos. But it is not yet perfect enough in solving this problem:

Although quite good!

7. What is shown in the photo?

Clarifai.com can determine what is shown in a photo. It is easy to recognize that in this picture there is an adult man with glasses and a beard:

Determines the dog breed:

But I have not yet learned to distinguish mulberries from other fruits:

8. Gemini or not?

Microsoft's entertainment service can determine whether people in photos are different or the same person.

Despite his years and glasses, he recognizes Steve Jobs:

And distinguishes him from Tim Cook:

9. Human emotions

And here is a joint project between Microsoft and Oxford University to analyze the emotions of a person in a photo.

The service easily determined that with almost 100% probability the woman experiences fear:

And he knows what a happy person looks like (“happiness”):