Lockdown and Window Control
Integrated ctypes, psutil, and PyAutoGUI to detect and close unauthorized windows, enforce allowed URLs inside QWebEngineView, and capture screenshots during suspicious events.
Real-Time Monitoring Pipeline
Combined MediaPipe FaceMesh (for eye openness, gaze, and head pose) and YOLOv5 (for phone detection) into a QThread-powered real-time processing pipeline.
Tuned focus scoring thresholds, penalty and recovery logic, and event-logging mechanisms to balance sensitivity while minimizing false positives.
Reporting and Data Visualization
Built dynamic PDF report generators using ReportLab, embedding focus-trend charts generated with Matplotlib and detailed penalty/event logs.
Leveraged Pandas for data aggregation and Matplotlib for live score visualization in both the UI and reports.
Concurrency and Performance Optimization
Managed intensive video processing tasks via QThread and Python’s threading module to maintain a responsive user interface.
Optimized model inference (resolution tuning, batched processing) and database writes to ensure compatibility even on low-spec lab computers.
Testing and Refinement
Conducted end-to-end testing across Windows setups, debugging edge cases such as webcam disconnections, rapid window switching, and unexpected system behaviors.
Polished the user experience with responsive design, clear error handling, and performance profiling.
Log in or sign up for Devpost to join the conversation.