PROJECT NOTE
SocialSync
Exploring how real-time visual feedback can make social cues easier to understand.
- ROLE
- Full-stack development lead
- CONTEXT
- Team assistive-technology prototype
- YEAR
- 2024
- TOOLS
- Python, OpenCV, TensorFlow, Electron, Flask, PyQt5
OVERVIEW
Project overview
A team desktop prototype exploring visual feedback for autistic people interpreting social cues. The original PyQt5 application connects camera-based expression classification to confidence-weighted feedback. The current Electron implementation sends camera frames to a local Flask/TensorFlow service and stabilizes classifier outputs over time. Clinical outcomes were not established by this project.
BUILD NOTES
My contributions
Combined face detection with a TensorFlow expression classification model.
Connected the current Electron camera interface to local model inference.
Stabilized classifier outputs with temporal smoothing and confidence-weighted recent detections.
INTERACTIVE PROJECT
Explore SocialSync.
Continue from the project note to the interactive browser experience.