Search by faces: how to find other photos of a person (Yandex and Google). Facial recognition software allows you to find any person on the network by taking a photo of them Face recognition
In recent times, facial recognition programs have been exclusively in the service of government agencies. They help identify a person based on facial features. Nowadays, the use of such software is most often seen in security systems and in commercial activities related to advertising and promotion of various services.
Also, some semblance of such online services and programs is available to ordinary users; they can be found on the Internet. An example is the Lenovo Veriface program. This is a proprietary built-in utility that allows you to unlock your device by analyzing your face.
How does face search work?
In some services, search by faces works on based on neural network technology. The search is based on certain basic features: image weight, resolution, color correction, exposure, resolution, etc. Nowadays, neural network technology is developing, in addition to the basic characteristics, the definition occurs using additional features: skin texture, shape of mouth, ears, hair color, eye location, etc.
To find a similar person, you must upload clear photos, otherwise the search will fail. It is also worth saying that social networks protect information about their users. Most social networks have built-in profile protection features that allow you to hide a person’s photos.
Facial recognition programs
Facial recognition systems (programs) from photos and videos use more advanced algorithms. Similar systems not publicly available, they are used in security services, to control and prevent emergency incidents in large cities and in crowded places.
Such software most often uses the Viola-Jones recognition algorithm. Using this method, the program recognizes faces even when rotated by 30 degrees. Method based on Haar characteristics, which are a set of black and white masks of various shapes and sizes. Masks are applied to the image and pixel brightness is calculated by adding, after which the difference in values is calculated.
Next, the system compares the results with the accumulated data, and having identified the face in the image, it continues to track it to select the optimal angle. After this, the reading process starts and the software finds reference points on the face. Next, the image is analyzed with the data that is in the databases. If they match, the person is identified.
FindFace
FindFace is a Russian facial recognition program, a powerful solution that will allow you to recognize a face in a photo. FindFace offers not only tools for ordinary users, but also for government corporations and businesses. With its help you can find a person registered on VKontakte.
From September 1, the public Find Face service will stop working; for some time, only those users who have purchased a PRO account will be able to use it. The authors will then abandon the public domain version entirely in favor of their commercial product.
Google Images
Google.Images search allows you to find identical photos based on search engine data, including searching for a person by face. More focused on search in foreign resources:
Yandex Pictures
A domestic analogue of the previous service, which will help you find a person by photo on the Internet and search by photo on social networks:
Pictriev
An English-language service that allows users to find a similar person and recognize a face from a photo online. First of all, it is focused on comparison with celebrities, so the results may not be entirely correct:
Yandex and Google allow you to search by faces and also find similar images.
Why is this necessary?
For example, this will help find other photos of a person from an existing photo.
I think the simplest thing is to find copies of the image in various formats, with additional elements, etc.
Let's take a photo of the famous actress Monica Bellucci and use it to test Yandex and Google.
Search engines have special operators that allow you to search by person.
Now we take a photo of a person from social networks, the result: 
Obviously, there are no other photos of this particular lady here, but there are plenty of relatively similar ones.
More nuances of Yandex image search are in the following video:
conclusions
| NEW: based on photos and other information. |
Finding copies of images is not difficult for search engines. However, if we are talking about searching by face for other photographs of this person, for now such a search is more relevant for people who have many photos on the Internet that are openly accessible for indexing.
Explore the article– here you can also find information about photo storage services, which can also be used to search for similar faces or specific people.
How to find a person without knowing their last name, first name, city, age? There are ONLY PHOTO. No, we will not use intelligence technology. There are online services that will easily allow anyone to do this for free. There can be many applications for this: finding random passers-by, love at first sight, old friends, determining the identity of a criminal, finding doubles, contacting the owner of lost documents.
Preparing the photo
The search accuracy will be higher if you prepare a photograph. Trim off excess, remove strangers, position your face straight. This can be done using any photo editor.
The simplest option on Windows is standard Paint.
As an example, let's take a photo of Mr. Bean from the FaceApp application. We want to find a woman like Mr. Bean.
- Launch Paint through Start - All Programs - Accessories, or find it through a search in Start.
- Open the photo. File - open. Or drag the file into the program window.
- Select the “Select” tool. We outline the face we will be looking for. Then press the crop button.
- Save the photo in jpg or png format: File - save as...
If the face is positioned unevenly, we additionally use the Rotate tool. Standard Paint can only rotate a photo by 90°.
VKontakte profile search
The FindFace service can still search for people from VKontakte.
- We go to the website https://findface.ru/ and click “Find identical ones”.
- To search, log into the site through your VKontakte account. If it is not there, register a new account on vk.com. The service will request access to your account, allow it.
- Upload the photo to the service and wait for the search to complete. According to the developers, the search accuracy is 70%.
The free account allows for 30 photo searches and 5 views of found profiles. If you need more, you will have to pay for a Premium account for 149 or 459 rubles per month. If you are going to constantly look for people, it is more profitable to pay for a year - there will be a discount. You can also get a free monthly premium by inviting 10 friends.
To increase the accuracy of your search, specify your age, city, and relationship status on VKontakte. Please be aware that many people have the wrong ages listed.
The most similar faces are at the top of the list. You can go to the VKontakte page, add a person to your likes list, or chat with him through the FindFace mobile application (for Android). Using the mobile application, you can take photos and immediately determine your identity. Imagine the surprise of a stranger when you guess his name, favorite movies, music. A good reason to get to know each other
A few examples:
The girl was taking photos for documents and forgot the flash drive in the photo studio:
The girl met a guy on the bus, but was afraid to approach:
We find a person using a photo from a passing bus:
“Girl, has anyone ever told you that you look like Vin Diesel?”
Search the entire Internet
If a person does not have a profile on VKontakte, let's try to find him through a search by image:
- Search by image from Google https://www.google.ru/imghp?hl=ru&tab=wi.
- Search by image from Yandex https://yandex.ru/images/.
Technologies
Now any person can be found on social networks just by photographing their face.
A new Russian application compares pictures with profile photos on the social network VKontakte and establishes his identity with an accuracy of 70 percent.
A new facial recognition app could end anonymity in public spaces.
Thus, a person can take a photo of a stranger on the street, upload it to the app and find your profile on the social network, while stores, advertisers and the police will be able to recognize your face in the crowd and follow you through social networks.
After its launch, the app quickly attracted more than 500,000 users and processed more than 3 million searches.
Russian facial recognition program from photos
The creators of the application are 26-year-old Artem Kukharenko and 29-year-old Alexander Kabakov.
Unlike other facial recognition programs, their algorithm allows for quick searches in large amounts of data.
The app gives you the best match for the face in the photo, along with 10 people it might resemble.
Artem Kabakov believes that this application can make a breakthrough in the dating industry: " If you see someone you like, you can take a photo of them, find their identity, and send a friend request".
Moreover, it also helps to find similar people. For example, you can upload a photo of a celebrity you like or your ex-lover and find 10 similar people by messaging them.
Photographer from St. Petersburg Evgeny Tsvetkov recently conducted a photo project " Your Face Is Big Data" ("Your face is a large database"), photographing 100 random people who were sitting next to him on the subway. Then, using the Findface application, he tried to find them on the VKontakte social network.
It turned out that it was quite easy to detect 60-70 percent of people aged 18 to 35, a little more difficult with older people. Thus, even though he took low-quality photographs, he was able to find out personal details of their lives, including hobbies, professions, relationship status and habits.
The creators of FindFace believe that such a program, among other things, will help in investigating crimes by identifying criminals on the street.
FindFace - Internet facial recognition program
FindFace is a service that helps you can find a VKontakte profile by photo. There are mobile applications for Android and iOS, as well as a web version.
The application allows you to do 30 free searches, and after that the service is paid.
The web version has a little more functions, and you can immediately see a person’s profile.
The program works best at recognizing photos taken in daylight on an average smartphone, but is 2-3 times less accurate if you're photographing strangers from afar. But by enlarging the image, efficiency increases again.
The development of technology leads to the fact that less and less information remains hidden, and end of anonymity.
Adviсe:
· Experts advise putting something on your face, such as a hood, mask or goggles, to fool the camera, or come to terms with the fact that in our society there is less and less room for personal space.
If you don't want any stranger to find out about you, it is best to delete all old photos from your account and leave a photo that will make it difficult for you to be recognized.
Is a computer application that can identify or verify a person from a digital image or video. One way to do this is to compare selected facial features with an image and options from a database.
Facial recognition (software) is commonly used in security systems, and can be compared to other biometric systems (eg fingerprint or eye recognition). Recently, it has also become popular as a commercial identification and advertising tool.
Some facial recognition algorithms identify facial features by extracting landmarks or objects from an image of the subject's face. For example, the algorithm may analyze the relative position, size and/or shape of the eyes, nose, cheekbones and jaw. This data is then used to find other images with matching parameters. Other algorithms normalize a gallery of facial images and then compress the facial information, retaining only the data in the image that is useful for facial recognition. The searched image is then compared with the available data. One of the earliest successful systems is based on template matching techniques applied to a set of characteristic facial features, providing a kind of compressed representation of appearance.
How does facial recognition software work?
It includes certain algorithms that can be divided into two main approaches:
- Geometric, which considers the distinctive or photometric features of the face.
- Statistical, which translates an image into values and compares them to patterns to eliminate variances.
3D recognition
A new trend created to achieve higher accuracy is 3D facial recognition. This method uses 3D sensors to collect information about the shape of the face. This information is then used to identify distinctive features such as the outline of the eye sockets, nose and chin.
One of the advantages of 3D facial recognition programs is that they are not affected by changes in lighting, unlike other options. This technology can also identify from a different range of viewing angles, including profile views. 3D data points significantly improve facial recognition accuracy. 3D research is being enhanced by the development of sophisticated sensors that improve the performance of 3D image capture. The sensors work by projecting structured light onto the face. Up to a dozen or more of these image sensors can be placed on a single CMOS chip - each capturing a different part of the spectrum.
However, even a perfect 3D matching method can be sensitive to facial expressions. For this purpose, a team of researchers at Technion applied tools from metric geometry to treat expressions as isometries. Vision Access then created its 3D facial recognition solution. The company was later acquired by Bioscrypt Inc., which developed a version of human facial recognition software known as 3D FastPass.
The new method is to introduce a way to capture a 3D image using three tracking cameras that point at different angles. One of them will point to the front side of the object, the second - from the side, the third - at an angle. They will all work together to be able to track the subject's face in real time and be able to identify it. It is believed that any program for facial recognition through a camera will soon be based on this technology.
Skin Texture Analysis
Another new trend uses visual skin detail that is captured in standard digital or scanned images. This technique, called skin texture analysis, turns the unique lines, patterns and spots visible in a person's skin into a mathematical space.
Tests have shown that with the addition of this technology, facial recognition efficiency can increase by 20-25%.
Thermal chambers
Another form of receiving facial recognition input is through the use of thermal imaging cameras. Thanks to this procedure, cameras will only detect the shape of the head, and will ignore objects such as glasses, hats or makeup. The problem with using thermal images for facial recognition is that the databases for it are limited.
Researchers are currently studying the use of this technology in real life and operational landscapes and at the same time creating a new database of thermal imaging images. The study uses low-sensitivity, low-resolution ferroelectric electrical sensors that are capable of long-wave thermal infrared (LWIR) imaging. The results show that fusion of LWIR and conventional imaging cameras has great results in outdoor probes. This combination can power a very powerful facial recognition program for the camera.
Mass use
While researchers are working on the latest technologies available only to specialists, developers of mass applications are also not standing still. After the creation of Google Glass was announced, there was a lot of noise around facial recognition and programs for this on the Internet. It is believed that this will open up a lot of opportunities for users not only to interact with each other, but also with various objects.
Today there is a large list of facial recognition interfaces available that you can use for your applications. The most common are the following.
Face Recognition Stephen
A development from Lambda Labs, which provides facial recognition based on the location of the eyes, the shape of the nose and mouth, and also carries out gender classification. Available on the developer's official website.
Face Detection
An online facial recognition computer program that is an ideal replacement for Face.com. Currently available free of charge.
Animetrics Face Recognition
The Animetrics Face Recognition API can be used to detect faces in photos. Information about facial features or landmarks is returned as coordinates in the image.
Skybiometry
To use it, you must have an application created in your SkyBiometry account. To create it, a simple registration is enough.
Face++
This application uses advanced technology and to provide 3 main services (detection, recognition and analysis). The program provides detection and analysis of Landmark (23 points), landmark (81 points), attributes: age, gender, glasses, race and so on.
FaceMark
It is a powerful facial recognition API. It finds 68 landmark points for the frontal face image and 35 for the profile one. FaceMark detects facial landmarks in an image specified by a URL or in an uploaded file, and produces the result as a JSON file containing a vector of facial landmarks and match points for each match found.
EmoVu by Eyeris
Intelligent emotion recognition software that allows cameras to read human micro-expressions, gender and age group. It is a real-time facial recognition program that works when users watch videos on their computers or mobile devices.
Rekognition.com
This is one of the best alternatives to Face.com. The fast, reliable and scalable rekognition engine can perform face detection, scanning, recognition and search. It can be automatically trained using images and tags on Facebook. Since it is social media oriented, it is the best facial recognition software for Android.
FaceRect
It is also a powerful and free face detection interface. It finds faces (both frontal and profile) in an image specified by URL or uploaded as a file, and can find multiple faces in a single photo, and outputs a response in JSON format. This loads the image with a bounding box for each face found.
