As students at the University of Waterloo, we are always open to opportunities in a broader community. I started to look for events near me using Facebook and realized that a lot of the events or information I learn is through Facebook groups. With these experiences in mind, I was inspired to create an application that posts relevant information on Facebook Groups. Hence, when groups are specifically created for certain reasons, they can receive relevant information in a smart, efficient, and automated way.
What it does
An image and text processing artifical intelligence PyTorch project to detect relevant ads for appropriate Facebook groups.
How we built it
Our application is built using neural networks, Google Cloud, Facebook Groups, and Flask is used for the web server.
Challenges we ran into
We were able to overcome the challenges we ran into by collaborating on our project and working towards learning new concepts. Being introduced to machine learning in terms of programming was an overall objective throughout this project. Saving and loading models to increase the efficiency of our application was an important step in order to be able to test our application faster. Using online courses and seeking advice from mentors, we were guided in the right direction to learn about optimizing machine learning applications.
Accomplishments that we're proud of
We are proud that our application has the power to actually post on groups automatically by analyzing an image uploaded to a web server. Seeing that our application can actually be used in Facebook through any smartphone shows that we can make a difference using PyTorch and Facebook's technologies.
What we learned
We learned about neural networks and the coding aspect of image recognition and artificial intelligence. It was also a great experience using APIs from Google and Facebook to deploy our application with cloud services.