SWVL: AI-Powered Face Tracking Camera Gimbal

This repository contains all files for SWVL, an AI-powered face tracking camera gimbal.
SWVL was developed as our senior Capstone project at the University of South Carolina by:

• Alex Anderson-McLeod
• Kuba Jerzmanowski
• Jagger Tanner
• Trevor Allison
• Michael Laitarovsky

Repository Structure

• The root directory mainly contains test files and can generally be ignored.
• The sw directory contains all host computer software, including:
  • Electron-based frontend
  • Python backend
• The hw directory holds all hardware-related files:
  • CAD files for the gimbal
  • Control PCB design
  • Firmware code running on the control board

Documentation

The file SWVL_Paper.pdf in the root directory provides a detailed description of the project, including objectives, methodology, and results.

Model Files

Due to file size restrictions, the trained AI model is not included in this repository. It is hosted separately on OneDrive.

Note: USC's OneDrive sharing settings currently restrict access to university users only.
If you are outside the university and need access to the model files, please contact me via email.

The latest model file is best_dev_loss0476_dsc.pth. Older versions are also available to review the development progress.

---

Feel free to reach out with questions or for further access!