StrangerSyncs

Detected & Retrieved Unique face from the database.

Inspiration

It all started with a small annoyance, forgetting where I left my stuffs and frustrated to look for it. But this little arrogance made me realize something deep. Imagine the number of people experiencing dementia such that in their everyday lives, they tend to forget who they've just met. We wanted to create a solution where the user does not have to forget them; it reminds them.

What it does

Our project uses facial recognition to provide details of the person you've met such as name, age and phone number. It is shown as a user interface (UI) to display the data of the person you've met and had a conversation with. If you meet a new person, you can add that person's details into the database.

How we built it

The prototype is designed to be as simple as possible to run the model. We used Python's OpenCV to detect the face closest to the camera, displaying a coloured square around the face. While the face is recognized within the frame, we can search (type S) that face using 128 dots projected onto the face, which this concept is similar to FaceID. Using that and the UUID, which is a unique ID, we can check if the face exists in the database. If it does exists, then a UI shows up beside the person's head, displaying the information of him/her taken from the data saved from the database. If it does not exists, then the UI shows asks to create this face. If I choose to create it, then a UI will appear where we can manually fill the details. So, we can stores all data into a collection of people saved in the database. Every time we meet the same person and search if that face exists in the database, then a UI appears based on the person's data. And the cycle goes on!

Challenges we ran into

Since we're using a prototype, we were limited with the resources to make an actual product. In addition to the project, we had to make sure we don't use too much space in the database, so we do things minimally. So, we couldn't implement the concepts that makes the prototype realistic such as voice recognition and implementing an AI agent to summarize a conversation since they may take up too much space.

Accomplishments that we're proud of

We're so proud that the facial recognition and detection work seamlessly!

What we learned

Through this project, we learned valuable insights in making the prototype such as facial recognition and facial detection working together, as well as facial recognition and Firebase. we learned how to crucially secure the databases to ensure we do not leak any API key.

What's next for StrangerSyncs

In the future, we aim refine and improve the prototype. This includes implementing it into custom AR glasses, adding AI-powered agents to implement voice recognition, summarizing their biography and previous conversations with the person. We also plan to gain any user feedback to improve the product.