Mirage

Inspiration

Suppose your are travelling to a tourist spot like 'Taj Mahal' and what to a perfect picture but there is no perfect moment to take the picture as it is very busy with people moving around you. Thus your supposed picture besides the Taj Mahal is always compromised (photobombed) by other people. This is where our application Mirage comes into the frame where it removes all the noise (moving people/objects) from picture to produce the Perfect Picture with just you and the Taj Mahal in it.

What it does

Mirage is a full Cognitive Photo application which detect your emotions and accordingly talks to you to proceed to the steps of taking the perfect picture and the generates auto captions for the photo. Firstly the application will detect you emotion and appropriately synthesis a speech based on your current emotion. Then come the photo cleaning part where a video (ie frames of pictures) of about 5-6 sec is recorded and fed to the system. The system builds a great noise free picture. This noise free picture is then fed to the server which uses 3 cognitive models to find the following: The person in the picture, his/her current emotion and the place where he/she is at that moment. Uses the output of all these we extrapolate the entire caption and tag it to the photo.. This final photo is noise free and auto captioned..

How we built it

We built this application using C++, OpenCv and Microsofts Cognitive APIs. This app is easily portable to mobile.

Potential

This kind of technology where we clean the image dynamically frame by frame is of a lot of value to current photo giants like Instagram and Snapchat whose user-base constantly seeks a perfect photo. Maybe even can rival them in the future with proper backing.