Inspiration

With classes and seminars moving to an online format, we realized there was a problem with the communication channel between professors and students.

During in person classes, for instance, we would be able to look around and gauge how other students are feeling. For topics that are confusing, we might be reassured that other students seemed just as confused or surprised by the new material, and more importantly, the professor would be able to assess this situation and slow down and re-explain some key points. Currently in Zoom, this channel of communication is not as readily available, so as a problem that we personally felt connected to, we were motivated to innovate and help create a solution.

What it does

Detects the general engagement of the class as well as allow the audience to anonymously input their current feedback.

Engagement detection is done by running facial detection and emotion recognition on the gallery mode of the Zoom call. This is presented on a concise web app where students can also input their feedback by clicking as well!

How we built it

The backend is coded primarily in Python. It communicates and stores data into our website and database that is hosted on AWS.

Challenges we ran into

  • One challenge was how to design and implement a dashboard that best visualizes and presents the important data. * Another challenge was how to encode different emotions into that visualization. For instance, how to weigh a happy emotion vs. say an angry one.
  • We also implemented the full tech stack, including the frontend development which is not our expertise.

Accomplishments that I'm proud of

  • Utilizing state of the art machine models
  • Built fully integrated pipeline
  • live auto-updated website

What we learned

  • Specifically, cache-invalidation for CSS / JS in chrome is not a fun bug to chase down. (Hint, hit shift+f5!!!)

What's next for Room Temperature

We see Room Temperature as a very easy to use website for any professor or speaker to utilize. The workflow is very simple! Professors can give out a room link for a class and would be able to gain meaningful feedback during their class. Afterwards, we would also present additional dashboard metrics that would indicate timestamps of when students were most confused or most engaged to help they improve their teaching!

Built With

Share this project:

Updates