Lookalike.me - service for people like you or your friends
The post was written at the request of a friend, who, together with a modest development team, made an excellent service - it’s not a pity to popiarite.
Hi, Habr!
I am one of the unlucky people who are being thrown by friends, girls and with whom people from the street do not want to meet, because I myself don’t come up to them, and for some reason they don’t approach me either.
')
A year ago, I decided to change. Having collected all my intent in my head, I sat down to draw the interface of the site for dating from photographs with a unique recognition system and finding similar faces. By the evening I had someone to discuss my sketch with, and today we are releasing a test version of the site lookalike.me - a service for finding people who look like you or your friends.
In this version, the main thing for us is to collect a base of photographs of individuals so that we can expand the functionality. For you - upload a portrait and see a person like it. It is noteworthy in the site that the results show only real people with whom it will be possible later to contact and chat, meet for tea or on the site connection.
Here, I just left me again. I go to the site, upload photos of my former and voila - new Ksyusha, I go.
reference
Lookalike is a service based on facial recognition technology designed to find and find people like each other by simply uploading photos (yours, friends, etc.). People around the world can find their counterparts and communicate with them. The service will be of interest to girls and guys who miss their ex.
In the face recognition program, elastic bunch graph mapping is used - the points responsible for the main parts of the face (nose, eyes, etc., about 50 pieces) are automatically located on the face. The similarity of persons is determined from the sum of the similarities of the neighborhoods of the corresponding points; these similarities are calculated using special formulas. It is based on the implementation of the algorithm from the Colorado State University - Face Identification Evaluation System, which we can use and improve, thanks to the CSU license. To search for faces in the photo, the openCV library is used.
UPDATE (by numerous requests ...)
Try service without registration
Your opinion is important to us.
The server does not maintain Habra. We apologize.
Hi, Habr!
I am one of the unlucky people who are being thrown by friends, girls and with whom people from the street do not want to meet, because I myself don’t come up to them, and for some reason they don’t approach me either.
')
A year ago, I decided to change. Having collected all my intent in my head, I sat down to draw the interface of the site for dating from photographs with a unique recognition system and finding similar faces. By the evening I had someone to discuss my sketch with, and today we are releasing a test version of the site lookalike.me - a service for finding people who look like you or your friends.
In this version, the main thing for us is to collect a base of photographs of individuals so that we can expand the functionality. For you - upload a portrait and see a person like it. It is noteworthy in the site that the results show only real people with whom it will be possible later to contact and chat, meet for tea or on the site connection.
Here, I just left me again. I go to the site, upload photos of my former and voila - new Ksyusha, I go.
reference
Lookalike is a service based on facial recognition technology designed to find and find people like each other by simply uploading photos (yours, friends, etc.). People around the world can find their counterparts and communicate with them. The service will be of interest to girls and guys who miss their ex.
In the face recognition program, elastic bunch graph mapping is used - the points responsible for the main parts of the face (nose, eyes, etc., about 50 pieces) are automatically located on the face. The similarity of persons is determined from the sum of the similarities of the neighborhoods of the corresponding points; these similarities are calculated using special formulas. It is based on the implementation of the algorithm from the Colorado State University - Face Identification Evaluation System, which we can use and improve, thanks to the CSU license. To search for faces in the photo, the openCV library is used.
UPDATE (by numerous requests ...)
Try service without registration
- Login: demo
- Password: demo
Your opinion is important to us.
The server does not maintain Habra. We apologize.
Source: https://habr.com/ru/post/82374/
All Articles