VaxCheck

Inspiration

Fighting COVID-19 has been a struggle over the past 2 years and we have reached near the end of the pandemic where vaccines are being administered. Although, there are a number of people who are not vaccinated which still makes gatherings risky and prevents our lives from going back to normal. The current system on checking whether a person is vaccinated or not is inefficient and can be improved. For example, we are asked to upload proof of vaccination in order to go to our school campus, but this is on a trust-based system and it does not prevent those who are not vaccinated from coming. We wanted to develop a website that solves this issue.

What it does

VaxCheck is a website that checks if you are vaccinated by using facial recognition and a photo ID database linked to your proof of vaccination. This can be used in public institutions such as universities and colleges so that it can match people using its photo ID database to see if they are vaccinated or not.

How we built it

VaxCheck is a class of very efficient CNN models with less than 1 million parameters that are specifically designed for high-accuracy real-time face verification on mobile and embedded devices. We begin by performing a basic examination of the weaknesses in mainstream mobile networks for face verification. Our specially constructed VaxCheck has effectively addressed this weakness. Under the identical experimental conditions, our VaxCheck outperforms other available APIs in terms of accuracy and speed by more than two times. Our single VaxCheck of 4.0MB size achieves 99.55 percent accuracy on LFW TensorFlow loss on the refined GoogleColab. On a GPU, the quickest VaxCheck has an actual inference time of 18 milliseconds. VaxCheck improves prior state-of-the-art mobile CNNs in terms of face verification efficiency. To make the website, we used HTML, CSS, and React.

Challenges we ran into

The biggest challenge was the time constraint. It is very hard to develop a complex idea like this with only one day. We used to have a fourth member but he ghosted us which was unfortunate. Despite these challenges, we were able to overcome them with hard work and determination.

Accomplishments that we're proud of

We are really proud of how our idea relates to making a contribution to stop the spread of COVID-19. The pandemic has hurt many people in a number of ways and making any step to prevent any more harm is very meaningful to us.

What's next for VaxCheck

Due to time constraints, we were not able to fully implement the database, so this would be our next step. Currently, the website only captures a photo and matches it with the photo ID, so we also want to make it so that it uses live footage. Once we complete that, we will contact institutions that use photo IDs such as universities, colleges, government facilities, and offices to see if we can implement our product into their COVID-19 regulation system.