We could have hundreds to thousands of photos in our phones. Group pictures, animal pictures, and, most importantly, selfies. Maybe we're looking for a particular selfie that we can't find, or maybe we want to delete them all. Whatever the reason, they can be hard to organize because of all the other photos we have. So here's our solution to this problem.

What it does

Our mobile app sends all the photos on the phone's gallery to our server to check whether or not the pictures are selfies using AWS and neural networks. It then displays all the selfies on the app with a count for the user. The app also gives the user an option to delete some or all of the selfies.

How we built it

For the front-end, we built a web app using HTML, CSS, and JavaScript., and an android mobile app using Java. For the back-end, we coded the servers using Django. For the selfie recognition, we wrote it in Python using libraries like scikit, pillow, and numpy.

Challenges we ran into

We were designing an app that pick All the selfies from your phone's gallery and deleted them to free up space. The important thing is that the app can pick out literally all selfies not just those identified by the cellphone (the cellphone can only identify those taken with front-facing camera).

Accomplishments that we're proud of

We at least finished a few key components to our project, including the face recognition (and selfie recognition), the back-end server, and the web app front-end.

What we learned

None of us had experience in AWS, back-end, or mobile development. During the hackathon, we all had our respective roles and we all spent that time learning about them.

What's next for Selfie Cleanse

Make it work.

Share this project: