Inspiration
Our inspiration was the technologies we wanted to learn at this event (Neural Networks, Front end, cloud storage, and how they all connect). We all wrote down what we wanted to learn and made a project that fit our needs very well, while serving a good purpose.
What it does
The idea was to train a neural network to recognize whether or not faces were smiling, which would dictate whether or not they were a subject in the picture. Then we would blur out unwanted faces, it's not a perfect solution, but its a start.
How we built it
The neural network was going to be run in python using tensorflow, and using OpenCV to modify the images. The plan for the frontend was a simple single or double page react app that would upload the image and then receive and display it back to the user. The plan for the backend was to use python and flask to take the image and pass it to the api to be scanned and send back the information that would go to our algorithm for sentiment analysis. After which our blur algorithm would then blur out the unwanted faces in the background.
Challenges we ran into
Obtaining a data set that was large enough to thoroughly train our neural network that would play nice with it as well. Unfortunately we were not able to get our network to work with the data set we had. We had a hard time understanding how to pass the images through the network itself to train it initially. We also ran into problems using the Google Cloud Vision API.
Accomplishments that we're proud of
We were able to get the selective blurring working, and we all did learn invaluable skills in back-end development that we can put to great use on future projects.
What we learned
We learned the basics of creating a model for machine learning in python using TensorFlow and python, and modifying images using OpenCV
What's next for Selfie Saver
Well the next step for Selfie Saver is to implement our own machine learning algorithm for face detection and sentiment analysis to cut out the API middle man.
Log in or sign up for Devpost to join the conversation.