In a curiosity-inducing machine learning project that relies on face recognition technology, Microsoft has released an age guessing website dubbed How-old.net which tries to guess the age of people through their photos, but alongside this ‘fun activity’, the site also slurps up metadata that could be used for targeted advertisements.
Corom Thompson and Santosh Balasubramanian, Engineers in Information Management and Machine Learning at Microsoft, set out to explore how Microsoft’s newly released Face detection API fared when put against photos of people by having the API predict the age and gender of any faces recognized in that picture.
Having an optimistic and rather lowly target of 50 users for a test, the duo sent out hundreds of emails inviting users to try out their website and check if Microsoft’s technology is able to guess their age. To their surprise instead of tends of users, their site garnered a whopping 35,000 users within few hours.
Out of the thousands of tests performed by the site, most of the tests picked up the sex of the subject correctly, but there were variations as far as age is concerned.
Though the APIs are still in development and the results aren’t always correct, the website does provide an insight into what Microsoft is working on as far as AI, Machine Learning and face detection technologies are concerned.
Metadata
Microsoft clearly states that it won’t be storing the image uploaded by users, there is a little mystery surrounding the metadata that gets captured by the site when a user uploads an image. In the blog post, the engineer duo, state that they got real-time insights into how people were using this tool.
The information that was passed to the site included User Agent string of the users of the web site, the latitude and longitude of location from where the picture was uploaded, the device type from which it was uploaded, among other things. Below is the JavaScript Object Notation metadata of one of the photographs a user was uploading:
[ { "event_datetime": "2015-04-27T01:48:41.5852923Z",
"user_id": "91539922310b4f468c3f76de08b15416", "session_id": "fbb8b522-6a2b-457b-bc86-62e286045452",
"submission_method": "Search",
"face": { "age": 23.0, "gender": "Female" },
"location_city": { "latitude": 47.6, "longitude": -122.3 },
"is_mobile_device": true, "browser_type": "Safari", "platform": "iOS", "mobile_device_model": "IPhone"
} ]
The duo agrees that they have collected all the metadata extracted in the JSON file and streamed to Event Hubs for further analysis.
Microsoft hasn’t indicated if any additional analysis was or is performed on the collected data and whether they are intending to use this information for targeted advertisements in future.
