Log in
Sign up

MyHeritage.com face recognition

MyHeritage.com Face Recognition runs in 3 steps:

The digital photo (or scanned photo print) that you provide, is loaded

Face detection technology is applied to automatically detect human faces in your photo (*)

Face recognition technology is applied to recognize the faces detected in the previous step
( * ) If faces are missed out in step 2, you can mark them manually using the Marker.
Recognizing faces is done by algorithms that compare the faces in your photo, with all faces previously known to MyHeritage.com Face Recognition, through photos and meta-data contributed by yourself and other users. So the more photos added to the system, the more powerful it becomes. If people in your photos are not recognized well, it is likely that MyHeritage.com has never encountered them before. By adding these photos to MyHeritage.com and annotating the people in the photo manually, MyHeritage.com will "learn" these faces and will be able to recognize them in future photos, even in different ages of the same person's life.
Note: the algorithms used by MyHeritage.com Face Recognition are likely to find relatives of people in your photo, due to the genetic-based facial similarities that exist between relatives. You can use this to form connections between people whom you never even knew were related.

Understanding the results


Master photo:
This is the photo that you provide us

Detected faces:
Each automatically detected face is displayed with a different colored border

Face recognition match:
Face recognition matches are displayed outside your Master photo, in cards that are described in detail below

Matching face: The face among the many thousands of faces in the growing MyHeritage.com photo collection, that most closely resembles the person in your photo

Matching person: The name of the person, whose face is the matching face

The similarity score: The face recognition similarity score between the face in your photo, and the matching face. Scores above 80% indicate a very likely match. The lower the score, the smaller the chances that the matching person is the same person as in your photo

More matches: Important! Use these buttons to navigate to additional matching faces. Matches are sorted by the similarity score, and often the correct matching face will not be the first one displayed but one of the others, so this is well worth using

Face gallery: Click this to see additional facial photos of the Matching person

Contact: Click this to contact the user who submitted the photo with the Matching face, to MyHeritage.com

More info: Click this for more information about the Matching person. This is typically displayed like this:
Any known information about the birth and death years and places of the Matching person is displayed, subject to privacy settings specified by the person who submitted the photo

Full size: Click this to view a full size image of the photo in which the Matching face was found
Additional instructions:

Marker Click this to manually mark faces in your Master photo

Zoom Click this to zoom in on any portion of your Master photo

Faces 1-4 of 13
If there are many face recognition matches, you can use these large arrows to navigate to additional cards

Moving the mouse over any card will draw lines that connect it to the face in your photo that it most closely matches

You can click any framed face in your photo to have its card come into view, if there are many face recognition matches
Hints & tips:

Face recognition works best if you can assign a year to the photo that you provide. This helps MyHeritage.com rule out people who were not alive when the photo was taken

When you submit photos to MyHeritage.com you can set their privacy level. This controls whether or not face recognition results run by other users on their photos, will include meta-data and faces from your photos

Face recognition shows matching faces without focusing on a specific gender. However, each card has a gender selector that allows you to restrict its results to a specific gender
 
Loading...
Loading...