AIGotU

Inspiration

After the pandemic, we have been online learning through ZOOM for about 8 months. In most classes, teachers do the roll call for the attendance record and also grading. For small classes, it seems ok and not take a too long time for this. However, for some classes, like Band, there are 100+ students in the same ZOOM and the director always has a headache about who is in the class and who is not. The roll call is not only taking a lot of time for teachers to match the students from a very small window on ZOOM but also impact the students' learning time in the class. This is the problem I want to resolve.

What it does

My solution is “AIGotU” – AI got you. This application can automatically capture the video stream and do the face detection to see if the student is already turn-on the video and already sit in front of the camera. Once the face is detected, the face recognition process starts to compare the input “Query Picture” and “Reference Pictures” from the database until finding the matched one. Then the student name will show on the screen and also update the information to the backend spreadsheet. So teacher can easily understand the attendance information of students just a glance.

How I built it

I use “face-api.js” as the kernel to build the website for the AIGotU application. Applying the face detection and face recognition functions from the API and modify the website and interface to show the recognition result and pictures. Then the result will be sent to the backend to generate the spreadsheet for teachers.

Challenges I ran into

The recognition is not so precise by using “SSD MobileNet V1” from the face-api.js. Sometimes the matching results are incorrect so we need to refine the database and query case. The best way it to enhance the algorithm but it’s no time to do it in a short period time of Hackathon.

Accomplishments that I'm proud of

The problem I found about the roll call for attendance from ZOOM and the solutions I want to provide to the teachers is what I am proud of. It can resolve the real issue we have every day during the pandemic to help both teachers and students in schools in the whole world. And more, This application can also be used on conferences, meetings, and any event using a virtual platform.

What I learned

This is the 1st time that I used the API to implement the AI application and I am amazed by the powerful and magical what AI can do. It’s so interesting and I want to do further analysis on vision computation research to understand it more.

What's next for AIGotU

Enhance the face recognition algorithm and import it to ZOOM

Built With

• api
• bootstrap
• css
• face-api.js
• html
• javascript