EmoSense
with Zeus Karl Pascual
Machine vision-assisted facial emotion recognition that helps guidance associates understand student emotional stability during assessment. A student answers a 25-item yes/no questionnaire while a camera reads their facial expressions. The system tallies the emotions across the session, sends the results to the OpenAI API, and shows counseling recommendations on a web portal.
I came up with the idea and wrote the system's code, shaped by our adviser's suggestions. My co-researcher Zeus did more of the written paper.
- 4.54group mean, Outstanding (ISO/IEC 25010)
- < 150 msper emotion classification
- OpenAIAPI writes the counseling recommendations
GPIO button starts and stops a session
- Raspberry Pi
- Python
- OpenCV
- TensorFlow Lite
- PHP / MySQL
- OpenAI API
- Prompt engineering