Campus Pulse

Main Dashboard of Campus Pulse
How We Utilized "Rekognition"

Inspiration

Have you ever walked into a library only to realize that there is not a single spot avaiable? Especially during midterm and final season, are you desperate to find a seat to sit down and lock in? To deal with these circumstances, we have come up with "Campus Pulse", where you can easily check the availability of each study space.

What it does

"Campus Pulse" allows students to easily check how busy campus libraries are floor by floor, view operating hours, and stay updated on important announcements, such as temporary closures.

How we built it

We built our website using React for the front end, which allows students to easily view library occupancy, operating hours, and announcements. For the backend, we used Python to integrate the AI detection system with AWS Rekognition, which processes camera footage to count the number of people in a space. The processed data is then sent to the database using DynamoDB, which is then fetched by the website and displayed in real time, allowing users to quickly check how busy a library is before visiting.

Challenges we ran into

After coming up with our idea, we encountered a major challenge: how to check library availability. We explored several methods to estimate how many people were inside a building and on each floor. Using Wi-Fi connections was one option since most people carry devices connected to the campus network, but this raised privacy concerns. We also considered tracking entries and exits using T-cards, but this would only work in certain libraries like Gerstein and Robarts, while many others do not require T-card access. Ultimately, we chose a camera-based detection system, where cameras integrated with AI-powered detection using AWS Rekognition count the number of individuals in a space and estimate occupancy by comparing the number of people to the available seats, with the data reflected on our website in real time.

Accomplishments that we're proud of

At first, we were not very familiar with AWS services, especially "How to integrate AWS within our code". However, from this Hackathon, we figured out ways to integrate them and successfully integrated, which we were very proud of. In addition, we were able to build a fully functional prototype within the time range.

What we learned

We were amused by the flexiblity and adaptability of AWS. Especially, when detecting number of individuals, we integrated "rekonition", AWS AI, which made this project way more efficient, without the need to train our own object detection model. We also learned how to integrate the cloud database that AWS provides into our project, using the database to store all occupancy data, and also create a public url to host the website.

What's next for Campus Pulse

Add/link library study room booking
Based on the information of occupancy we collected over a long period of time, we could theoretically predict the upcoming occupancy for each library/study area