Who is in the picture?
Face recognition is nothing unusual. Modern digital cameras are able to recognize faces when they take a picture. Some better models are also equiped with smile recognition, to take the picture at the perfect moment. And social media platforms (such as facebook) use face recognition to find the faces of your friends. You can mark your friends and let the social network knows with whom you have been there.
Artificial intelligence algorithms go one steh further: They are also able to analyze the face in the picture. Let me show you two examples:
Amazon Rekognition
AWS Rekognition is a deep learning-based image recongition. It is part of Amazon Web Services Artificial Intelligence services.
It detects faces, the age of the person, if someone is wearing eyeglassses, etc… The white dots on the picture mark the eyes, the nose and the corners of the mouth.
Watson Visual Recognition
Watson is part of IBM’s Bluemix cloud service. It currently offers 14 different AI cloud services. Visual Recognition is one of these services.
Watson provides a very poerful API. You can use it to analyze objects in a picture, categorize pictures based on samples and it offers also a very powerful face recognition. The coolest feature, at least in my opinion, is that Watson can also detect the identity of the person in the picture.
I used Watsons API and showed “him” the picture above. It is a picture of Ellen DeGeneres English Wikipedia article. Watson detected three people: Portia de Rossi (with a probability of 73%), Ellen DeGeneres (with a probability of 82%) and the unkown, sunglasses-wearing person in the background. The API responds with the following JSON document:
{
“images”: [
{
“faces”: [
{
“age”: {
“max”: 44,
“min”: 35,
“score”: 0.418274
},
“face_location”: {
“height”: 125,
“left”: 123,
“top”: 61,
“width”: 111
},
“gender”: {
“gender”: “FEMALE”,
“score”: 0.993307
},
“identity”: {
“name”: “Portia de Rossi”,
“score”: 0.731059,
“type_hierarchy”: “/people/celebrities/portia de rossi”
}
},
{
“age”: {
“max”: 64,
“min”: 55,
“score”: 0.371479
},
“face_location”: {
“height”: 131,
“left”: 250,
“top”: 107,
“width”: 126
},
“gender”: {
“gender”: “FEMALE”,
“score”: 0.970688
},
“identity”: {
“name”: “Ellen DeGeneres”,
“score”: 0.817574,
“type_hierarchy”: “/people/celebrities/ellen degeneres”
}
},
{
“age”: {
“min”: 65,
“score”: 0.827884
},
“face_location”: {
“height”: 81,
“left”: 460,
“top”: 381,
“width”: 70
},
“gender”: {
“gender”: “FEMALE”,
“score”: 0.0
}
}
],
“resolved_url”: “https://upload.wikimedia.org/wikipedia/commons/5/58/PortiadeRossiEllenDeGeneresHWOFSept2012.jpg”,
“source_url”: “https://upload.wikimedia.org/wikipedia/commons/5/58/PortiadeRossiEllenDeGeneresHWOFSept2012.jpg”
}
],
“images_processed”: 1
}These are only two examples of powerful AI-based face recognition algorithms. You can use their API to implement them in your own app or website.
Amazon charges 0.001 US$ per processed picture, IBM 0.002 US$ per tagging and 0.004 US$ per face recognition.
