Put a Face to It

Get to know someone and identify faces and names seamlessly.

Comment

Inspiration

One of the common situations which we've almost always encounter is the problem of being unable to recognise people whom we've just met. So what if we had a tool which tells us immediately who the person is?

What it does

It recognises an individual's face, put a name to it and allows users to connect with the identified person through Facebook.

How we built it

Put a Face to It integrates Microsoft's Cognitive Services APIs to automatically detect faces using the android phone camera and tells you who the person you are pointing the camera is. An extension of this function is the ability to track mutual friends with the identified person through Facebook. This helps an individual to easily identify and be in touch with the person, strike a conversation with the person!

Challenges we ran into

One of the problems we ran into was that the Clarifai API didn't work as well as we thought in identifying faces. The accuracy of detection was not what we'd expected initially. In the end, we tried out Microsoft's API and trained the set of faces again.

Initially, we also wanted to scrape friends' facebook data through login through Python, but then we realised that it might not be plausible as the friends data had to be stored in another server in the background while the Put a Face to It runs on the android phone. Hence, we decided to go ahead with allowing the user to login to Facebook and obtain mutual friends with the identified individual via permissions.

Accomplishments that we're proud of

Making the face detection work on android and syncing it to display mutual friends. Debugging.

What we learned

We learned that it is not viable to scrape data prior to running the app as it requires a server to store the data prior to accessing it. There are several face detection APIs out there and it would be good to explore them before deciding on which one to use.

What's next for Put a Face to It

Put a Face to It aims to implement a feature whereby the user can learn more about the identified person's interests, hobbies, career, education and calendar to seamlessly keep in touch.

Built With

Share this project:

Updates