while(i<= 6°) { friend = friendOfFriend }

Direct relation
A shortest path to find your mutual connection
Our Team !!!!

Inspiration

You meet a stranger on street but you seem to know them from somewhere. Won't it be awkward if you don't remember their name? - This AR application will help you avoid this situation

You are walking down a street and feel you are alone and in a foreign land, but really are you? There are people related to you by some mutual connections everywhere.

link

What it does

We use Samsung's GearVR to access the augmented reality to identify in real-time how the person is related to you. It also shows the shortest path of your connection (example : friend of a mutual friend). It scans the environment and calls the machine learning API for facial recognition. Which maps the person to the database collected from popular social networks like LinkedIn, Instagram, Facebook to say how they are related to you.

How we built it

Use used the gearvr API's to write an android code which captures the images and calls the backend java spring code for image conversion. The java spring backend internally calls the python machine learning API's to recognizes the faces and query a mongodb database to identify which user is this and how its related to the user who submitted the query. We are currently using BFS for traversing the nodes in the database to identify the path between two users

Challenges we ran into

One of the major challenge was the exploration of uncharted territories in GearVR api due to lack of documentation and also machine learning API for facial recognition. Another issue we faced was conversion of a steam of bytes from android to a image suitable for recognition. Implementing the DLIB cross platform software library for image processing and facial recognition.

Accomplishments that we're proud of

Implementing GearVR API. Machine Learning API for facial recognition. Constructing a middleware in java spring which communicated between the AR and machine learning API. Querying MongoDB to retrieve appropriate results

What we learned

The exploration of GearVR API's and choosing the right gear for implementing the augmented reality. The machine learning concepts of how images as a byte stream can be used to recognize facial features. Optimization of queries in MongoDB. How to collaborate with team of different technology stack and integrate our work.

What's next for while(i<= 6°) { friend = friendOfFriend }

The concept can be used for natural disaster recovery where the people can find their relatives and make sure they are okay in case of emergencies. It brings solace and peace of mind. This concept can also be used to find overcome and prevent criminal activities by police authorities using it to recognize threats. Security surveillance in case of crowded places.

Built With

Submitted to

Revolution UC

Created by

I worked on android gear VR. The VR/AR domain was quite new to me and it took some time to figure out the API's and functionality. I was successfully able to use the AR and render text information on the display.

Srinivas Kulkarni
I worked on the middle ware, where I exposed the RESTful API, which our AR client calls to identify the connections between the client and the person he's seeing.
I also wrote the shortest path/ node connection logic in the software.
I deployed the application on AWS cloud, and integrated various modules together. My technology stack included, java, J2EE, spring, springboot, maven, AWS, linux, webservices.

Abhimanyu Chadha
Adithya Kashyap
Khushang Mehta