TeachWell

Inspiration

Since secondary school, we are asked to complete a teacher performance survey at the end of every year. Though we all wanted to better our classes and teachers, we often blanked on which parts of the class we enjoyed or disliked. Frustrated with the inefficiency of these surveys, we decided to create an class optimisation solution that correlates engagement and interest with teaching style in real-time.

What it does

Uses facial recognition to analyze and track student response to class material.

• Uses Microsoft Azure Face technology to analyze facial expressions of students (anger, contempt, disgust, fear, happiness, sadness, surprise) and calculate dominant emotion
• Generates realtime dynamic graphing of the overall emotion in class.
• Generates downloadable report that displays how different emotions varied with time
• Also allows user to examine how each individual student responded to different materials
• Eliminates random personnel processing. Does not recognize faces that have not been inputted

How we built it

1. Use postman to collect images detected by the cisco Mareki camera
2. Send images to Microsoft Azure Face AI endpoint
3. Using the Python client SDK provided by Microsoft, process image data - includes identification of dominant emotion, face recognition and tagging, looping snapshot processing every 15 seconds, generate moving average
4. Send to frontend developed using react.js.

Challenges I ran into

Accomplishments that we are proud of

• Successfully integrating Microsoft Azure Face AI into data processing through the client library with Python
• Successfully

What we learned

What's next for TeachWell

• Add additional functionality such as allowing the teacher to add in time-stamps corresponding to different sections of the lecture
• integrate lesson plan uploading - smart processing of lesson plan to automatically generate time-stamp
• integration into online classroom platform - e.g. udemy. Analyzes images passed through webcam
• Eliminating the influence of personal factors in student response using negative feedback loop, compares student response to overall class response to see if there is complete disassociation with class engagement

Built With

• javascript
• mareki
• microsoft-azure-face
• python